# Conversational AI

Conversational AI is technology that enables software to process and respond to human voice- or text-based conversations using natural language processing and machine learning.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## When Should a Voice Agent Speak? End-of-Turn Detection in Turkish

DevFeed: [When Should a Voice Agent Speak? End-of-Turn Detection in Turkish](<https://devfeed.tech/articles/when-should-a-voice-agent-speak-end-of-turn-detection-in-turkish-145522.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25489>)

Author: Ahmet Tu\\u{g}rul Bayrak, Fatma Nur Korkmaz, Mustafa Serta\\c{c} T\\"urkel, Nur Ya\\u{g}mur Tuncer

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [speech-to-speech](<https://devfeed.tech/topics/speech-to-speech.md>), [Whisper](<https://devfeed.tech/topics/whisper.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [detection](<https://devfeed.tech/tags/detection.md>), [end-of-turn-detection](<https://devfeed.tech/tags/end-of-turn-detection.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [features](<https://devfeed.tech/tags/features.md>), [fine](<https://devfeed.tech/tags/fine.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [latency](<https://devfeed.tech/tags/latency.md>), [logistic-regression](<https://devfeed.tech/tags/logistic-regression.md>), [model](<https://devfeed.tech/tags/model.md>), [recording](<https://devfeed.tech/tags/recording.md>), [recordings](<https://devfeed.tech/tags/recordings.md>), [response](<https://devfeed.tech/tags/response.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [speaker](<https://devfeed.tech/tags/speaker.md>), [study](<https://devfeed.tech/tags/study.md>), [tests](<https://devfeed.tech/tags/tests.md>), [text](<https://devfeed.tech/tags/text.md>), [voice](<https://devfeed.tech/tags/voice.md>)

### AI overview

In Turkish dyadic video calls, a fine-tuned Turkish TurnGPT model distinguished turn ends from mid-utterance pauses better than a zero-shot Turkish language model and a deployed multilingual detector, and was best calibrated among the three text systems. Current-speaker voice-activity timing was similarly informative; combining timing with the fine-tuned text score produced the strongest timing-based results, including against combinations using either text baseline with logistic regression. In a latency simulation, adding the text score to a silence timeout reduced mean response delay while keeping the interruption rate unchanged. The fine-tuned model also retained its advantage over the zero-shot model on automatic Whisper transcripts. Evaluation used recording-disjoint and speaker-disjoint splits.

### Source excerpt

arXiv:2611.25489v1 Announce Type: new Abstract: A voice agent that waits for a fixed silence before replying either interrupts users who pause mid-utterance or responds late to users who have finished. This study examines the decision at the onset of silence, using turn ends and turn-internal pauses extracted from the Real-TurnTurk corpus of Turkish dyadic video calls. A fine-tuned Turkish TurnGPT model, a zero-shot Turkish language model and a deployed multilingual detector are compared, together with voice-activity timing and facial features, under recording-disjoint and speaker-disjoint evaluation with paired cluster bootstrap tests. On held-out recordings, the fine-tuned model separates turn ends from pauses better than both text baselines and is the best calibrated of the three. Timing features of the current speaker are similarly informative, and their combination with the fine-tuned text score gives the best timing-based results; with logistic regression, this combination also outperforms the same combination with either text baseline. In a latency simulation, adding the text score to a silence timeout shortens the mean response delay at the same interruption rate. On automatic Whisper transcripts, the fine-tuned model retains its advantage over the zero-shot model.

## A Quasi-experimental Study of Conversational AI-supported Engagement in Large CS Courses

DevFeed: [A Quasi-experimental Study of Conversational AI-supported Engagement in Large CS Courses](<https://devfeed.tech/articles/a-quasi-experimental-study-of-conversational-ai-supported-engagement-in-large-cs-courses-145304.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2506.18941>)

Author: Neha Rani, Sharan Majumder, Ishan Bhardwaj, Pedro Guillermo Feijoo Garcia

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [chatgpt](<https://devfeed.tech/tags/chatgpt.md>), [computer-science](<https://devfeed.tech/tags/computer-science.md>), [conversational-ai](<https://devfeed.tech/tags/conversational-ai.md>), [courses](<https://devfeed.tech/tags/courses.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cy](<https://devfeed.tech/tags/cs-cy.md>), [cs-et](<https://devfeed.tech/tags/cs-et.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [engagement](<https://devfeed.tech/tags/engagement.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [large](<https://devfeed.tech/tags/large.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [study](<https://devfeed.tech/tags/study.md>)

### AI overview

In a quasi-experimental within-subject study in a large Software Engineering course at a public U.S. university, students took part in both an in-class discussion using ChatGPT and a peer discussion without it. Researchers assessed CAI interactions and students' self-reported engagement. The CAI activity was associated with higher engagement with the learning material, higher self-reported learning, and significantly greater perceived usefulness for learning. The abstract presents CAI as having potential to support engagement in large classes; it does not establish effects beyond this course and study design.

### Source excerpt

arXiv:2506.18941v2 Announce Type: replace-cross Abstract: Lucrative career prospects and creative opportunities often attract students to enroll in computer science courses, resulting in large class sizes. A common challenge in large classrooms is the lack of engagement between students and both the instructor and the learning material. Recent advances in Large Language Models (LLMs) present an opportunity to leverage Conversational Artificial Intelligence (CAI) to address this challenge. To examine the potential of CAI to support engagement with learning material, we designed an in class activity in a large Software Engineering course in which students interacted with a CAI tool. Using a quasi-experimental, within-subject study design in a real classroom at a large public United States University. We compared student engagement during an activity that involved the use of CAI for discussion with an activity that did not (peer discussion). For our analysis, we collected data on students interactions with the CAI tool and their self-reported responses on multiple engagement scales. Results show higher student engagement with the learning material in the activity involving CAI discussion as opposed to without (peer discussion). Higher self-reported learning was also observed in the activity with CAI as compared to without. Students also reported significantly more usefulness for learning in the CAI condition. Overall, our results indicate that CAI (ChatGPT) has the potential to support engagement with learning content during in-class activities in large class sizes.

## Meta's AI agent Muse is mixing up conversations and exposing private data, report says

DevFeed: [Meta's AI agent Muse is mixing up conversations and exposing private data, report says](<https://devfeed.tech/articles/meta-s-ai-agent-muse-is-mixing-up-conversations-and-exposing-private-data-report-says-143920.md>)

Original publisher: [Read original article](<https://techcrunch.com/2026/11/19/metas-ai-agent-muse-is-mixing-up-conversations-and-exposing-private-data-report-says/>)

Author: Aisha Malik

Published: 2026-11-19T17:38:40Z

Content type: news

Language: en

Sources: [TechCrunch | AI News & Artificial Intelligence](<https://devfeed.tech/sources/techcrunch-ai-news-artificial-intelligence.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [sensitive data](<https://devfeed.tech/topics/sensitive-data.md>)

Tags: [account](<https://devfeed.tech/tags/account.md>), [advice](<https://devfeed.tech/tags/advice.md>), [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [app](<https://devfeed.tech/tags/app.md>), [apps](<https://devfeed.tech/tags/apps.md>), [bank](<https://devfeed.tech/tags/bank.md>), [career](<https://devfeed.tech/tags/career.md>), [career-advice](<https://devfeed.tech/tags/career-advice.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [data](<https://devfeed.tech/tags/data.md>), [employees](<https://devfeed.tech/tags/employees.md>), [in-brief](<https://devfeed.tech/tags/in-brief.md>), [messages](<https://devfeed.tech/tags/messages.md>), [meta](<https://devfeed.tech/tags/meta.md>), [mixing](<https://devfeed.tech/tags/mixing.md>), [muse](<https://devfeed.tech/tags/muse.md>), [orders](<https://devfeed.tech/tags/orders.md>), [private](<https://devfeed.tech/tags/private.md>), [product](<https://devfeed.tech/tags/product.md>)

### AI overview

Meta employees reportedly raised concerns that its AI agent Muse mixed up conversations across accounts and exposed private information. Examples in internal posts included unrequested childcare and career messages, product orders, bank balances, and a reminder about a Signal message that one employee said she had not authorized Muse to access. The account is based on a Reuters report of internal posts; Meta had not immediately responded to TechCrunch's request for comment. The excerpt also says Muse launched in September and had more than 10 million downloads.

### Source excerpt

In some of these instances, employees said Muse surfaced messages about childcare requests they never made, career advice they never asked for, product orders they never placed, and bank account balances that didn't belong to them.

## Conversational AI Testing: Types, Workflow, Challenges & Best Practices

DevFeed: [Conversational AI Testing: Types, Workflow, Challenges & Best Practices](<https://devfeed.tech/articles/conversational-ai-testing-types-workflow-challenges-best-practices-145968.md>)

Original publisher: [Read original article](<https://testgrid.io/blog/conversational-ai-testing/>)

Author: Tejas Patel

Published: 2026-11-19T10:59:06Z

Content type: article

Language: en

Sources: [TestGrid](<https://devfeed.tech/sources/testgrid-blog.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [Performance Testing](<https://devfeed.tech/topics/performance-testing.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Chat Bot](<https://devfeed.tech/topics/chatbot.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [access-controls](<https://devfeed.tech/tags/access-controls.md>), [actions](<https://devfeed.tech/tags/actions.md>), [adversarial](<https://devfeed.tech/tags/adversarial.md>), [adversarial-testing](<https://devfeed.tech/tags/adversarial-testing.md>), [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-testing](<https://devfeed.tech/tags/ai-testing.md>), [api](<https://devfeed.tech/tags/api.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [call](<https://devfeed.tech/tags/call.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [conversational-ai](<https://devfeed.tech/tags/conversational-ai.md>), [conversational-ai-testing](<https://devfeed.tech/tags/conversational-ai-testing.md>), [customer](<https://devfeed.tech/tags/customer.md>), [order](<https://devfeed.tech/tags/order.md>), [request](<https://devfeed.tech/tags/request.md>), [software-testing](<https://devfeed.tech/tags/software-testing.md>), [test-automation](<https://devfeed.tech/tags/test-automation.md>), [testing](<https://devfeed.tech/tags/testing.md>), [types](<https://devfeed.tech/tags/types.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

Conversational AI testing must evaluate meaning and outcomes across the full interaction stack, since responses can vary and fluent answers may still be fabricated or trigger incorrect actions. The source recommends testing intent and entity recognition, context retention, grounded accuracy, task and tool execution, safety and fallback, performance, security, and voice behavior where applicable. A workflow is to define objectives and pass criteria, design standard and edge-case scenarios, prepare varied and privacy-conscious data, run tests through components and end-to-end journeys, assess task completion, relevance, groundedness, tool success and latency, then investigate failures and rerun regression tests. Exact-string checks suit deterministic behavior, while open-ended answers need semantic or rubric-based evaluation, potentially with calibrated LLM judges and human review for subjective or high-risk cases. TestGrid supplies device, browser, audio-routing, network and execution capabilities; the test framework or evaluator judges response quality.

### Source excerpt

Traditional software testing loves an answer key. Click this button. Expect this screen. Send this request. Expect 200 OK. Then conversational AI walked in and decided that five different answers could all be correct, the sixth could sound correct while being completely fabricated, and the seventh might call an API and cancel your customer's order. [...]

## Risk-Calibrated Intent Routing for Conversational AI via Signed Lexical Evidence

DevFeed: [Risk-Calibrated Intent Routing for Conversational AI via Signed Lexical Evidence](<https://devfeed.tech/articles/risk-calibrated-intent-routing-for-conversational-ai-via-signed-lexical-evidence-142570.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.00262>)

Author: Yezhou Cheng, Zehua Yang, Bojun Lin

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Calibrated uncertainty](<https://devfeed.tech/topics/calibrated-uncertainty.md>)

Tags: [confidence](<https://devfeed.tech/tags/confidence.md>), [conversational-ai](<https://devfeed.tech/tags/conversational-ai.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [lexical](<https://devfeed.tech/tags/lexical.md>), [risk](<https://devfeed.tech/tags/risk.md>), [routing](<https://devfeed.tech/tags/routing.md>)

### AI overview

The method combines the gap between a sentence classifier's top two intent scores with signed lexical evidence from words and short character patterns, then uses separate calibration data and a statistical upper bound on error to set an acceptance threshold. Across ten runs per dataset, it reduced risk-coverage area versus a confidence model using classifier scores alone by 15.8% on BANKING77, 15.1% on CLINC150, and 11.8% on HWU64. At a 5% error target, accepted-request coverage increased by 1.83 percentage points on BANKING77 and 5.14 on HWU64; CLINC150 already had near-full coverage. At a 2% target, calibration found an accepting rule in all 30 runs.

### Source excerpt

arXiv:2610.00262v2 Announce Type: replace-cross Abstract: Conversational AI has become widely used as advances in language models let people find information, get writing help, and ask for practical guidance in their own words. When an assistant handles specific tasks, it must infer what the user wants and decide whether to act on that prediction. A sentence classifier's confidence can miss clues in the wording. We combine the gap between the classifier's two highest scores with signed lexical evidence, a score based on words and short character patterns. The score is positive when those patterns favor the predicted intent and negative when they favor another. We use separate calibration data and a statistical upper bound on the error rate to set the minimum confidence required to accept a request. Across ten runs per dataset, our method reduces the area under the risk-coverage curve, which summarizes error as more requests are accepted, by 15.8%, 15.1%, and 11.8% on BANKING77, CLINC150, and HWU64, respectively, compared with a confidence model that uses only sentence classifier scores. At a 5% error target, the share of accepted requests rises by 1.83 percentage points on BANKING77 and 5.14 points on HWU64. CLINC150 already has near-full coverage at that target. At a 2% target, calibration finds a rule that accepts some requests in all 30 runs. Our contribution is a two-feature confidence model that improves routing by measuring how strongly the wording supports or challenges the original intent prediction.

## Coinbase blocks $4.7 million in transfers to Elon Musk deepfake scam

DevFeed: [Coinbase blocks $4.7 million in transfers to Elon Musk deepfake scam](<https://devfeed.tech/articles/coinbase-blocks-4-7-million-in-transfers-to-elon-musk-deepfake-scam-141245.md>)

Original publisher: [Read original article](<https://www.bleepingcomputer.com/news/security/coinbase-blocks-47-million-in-transfers-to-elon-musk-deepfake-scam/>)

Author: Lawrence Abrams

Published: 2026-11-18T20:49:32Z

Content type: news

Language: en

Sources: [BleepingComputer](<https://devfeed.tech/sources/bleepingcomputer.md>)

Topics: [Cryptocurrency](<https://devfeed.tech/topics/cryptocurrency.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [coinbase](<https://devfeed.tech/topics/coinbase.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-powered](<https://devfeed.tech/tags/ai-powered.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [campaign](<https://devfeed.tech/tags/campaign.md>), [coinbase](<https://devfeed.tech/tags/coinbase.md>), [computer-help](<https://devfeed.tech/tags/computer-help.md>), [computer-security](<https://devfeed.tech/tags/computer-security.md>), [computers](<https://devfeed.tech/tags/computers.md>), [crypto](<https://devfeed.tech/tags/crypto.md>), [cryptocurrency](<https://devfeed.tech/tags/cryptocurrency.md>), [deepfake](<https://devfeed.tech/tags/deepfake.md>), [detection](<https://devfeed.tech/tags/detection.md>), [elon-musk](<https://devfeed.tech/tags/elon-musk.md>), [fraud](<https://devfeed.tech/tags/fraud.md>), [fraud-prevention](<https://devfeed.tech/tags/fraud-prevention.md>), [giveaway](<https://devfeed.tech/tags/giveaway.md>), [impersonation](<https://devfeed.tech/tags/impersonation.md>), [infosec](<https://devfeed.tech/tags/infosec.md>), [infosec-computer-security](<https://devfeed.tech/tags/infosec-computer-security.md>), [intervene](<https://devfeed.tech/tags/intervene.md>), [linux](<https://devfeed.tech/tags/linux.md>), [livestream](<https://devfeed.tech/tags/livestream.md>), [mac](<https://devfeed.tech/tags/mac.md>), [malware](<https://devfeed.tech/tags/malware.md>), [malware-removal](<https://devfeed.tech/tags/malware-removal.md>), [scam](<https://devfeed.tech/tags/scam.md>), [security](<https://devfeed.tech/tags/security.md>), [spyware](<https://devfeed.tech/tags/spyware.md>), [support](<https://devfeed.tech/tags/support.md>), [tech-support](<https://devfeed.tech/tags/tech-support.md>), [technical-support](<https://devfeed.tech/tags/technical-support.md>), [tesla](<https://devfeed.tech/tags/tesla.md>), [virus](<https://devfeed.tech/tags/virus.md>), [virus-removal](<https://devfeed.tech/tags/virus-removal.md>), [windows](<https://devfeed.tech/tags/windows.md>)

### AI overview

Coinbase says its real-time machine learning system blocked about 91% of attempted transfers--$4.7 million--in a roughly two-hour fake giveaway livestream impersonating Elon Musk. Hundreds of customers tried to send cryptocurrency after a deepfake video promised to double their deposits. Because customers authorized the transfers themselves, the system looked for giveaway and brand-impersonation patterns, including newly created wallets, promises to multiply funds, and event-related time pressure. When it flagged a transfer, an AI agent explained scam indicators and asked questions; confidence scores could trigger additional safeguards. Coinbase says the agent recognized techniques from a campaign it had not previously encountered and that intervention was automatic. The excerpt does not establish how much cryptocurrency reached the scammers; Coinbase had not responded to the publication's inquiry.

### Source excerpt

Coinbase says it blocked $4.7 million in cryptocurrency transfers to scammers promoting a fake giveaway using an AI-generated deepfake of Elon Musk during a Tesla livestream that attracted more than 500,000 viewers. [...]

## How Stretto built an AI voice agent with Amazon Nova 2 Sonic

DevFeed: [How Stretto built an AI voice agent with Amazon Nova 2 Sonic](<https://devfeed.tech/articles/how-stretto-built-an-ai-voice-agent-with-amazon-nova-2-sonic-141098.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/machine-learning/how-stretto-built-an-ai-voice-agent-with-amazon-nova-2-sonic/>)

Author: Mariah Miller

Published: 2026-11-18T18:58:42Z

Content type: article

Language: en

Sources: [Artificial Intelligence](<https://devfeed.tech/sources/artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [speech-to-speech](<https://devfeed.tech/topics/speech-to-speech.md>), [Amazon Web Services](<https://devfeed.tech/topics/aws.md>), [Amazon Aurora](<https://devfeed.tech/topics/amazon-aurora.md>)

Tags: [advanced-300](<https://devfeed.tech/tags/advanced-300.md>), [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-voice](<https://devfeed.tech/tags/ai-voice.md>), [ai-voice-agent](<https://devfeed.tech/tags/ai-voice-agent.md>), [amazon-aurora](<https://devfeed.tech/tags/amazon-aurora.md>), [amazon-bedrock](<https://devfeed.tech/tags/amazon-bedrock.md>), [amazon-cloudwatch](<https://devfeed.tech/tags/amazon-cloudwatch.md>), [amazon-connect](<https://devfeed.tech/tags/amazon-connect.md>), [amazon-elastic-container-service](<https://devfeed.tech/tags/amazon-elastic-container-service.md>), [amazon-nova](<https://devfeed.tech/tags/amazon-nova.md>), [amazon-sqs](<https://devfeed.tech/tags/amazon-sqs.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [aws](<https://devfeed.tech/tags/aws.md>), [bankruptcy](<https://devfeed.tech/tags/bankruptcy.md>), [customer-solutions](<https://devfeed.tech/tags/customer-solutions.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [interviews](<https://devfeed.tech/tags/interviews.md>), [lead](<https://devfeed.tech/tags/lead.md>), [phone](<https://devfeed.tech/tags/phone.md>), [production](<https://devfeed.tech/tags/production.md>)

### AI overview

Stretto replaced Bankruptcy.com's short web form with Tara, a phone agent that gathers bankruptcy intake details conversationally, then uses attorney-defined rules to sort leads for attorney review. Amazon Nova 2 Sonic handles speech input and output in one model; Claude 3 Haiku extracts typed fields from transcripts, which are stored in Aurora. Twilio streams calls to a voice bridge on ECS Fargate. Because full interviews can exceed one streaming session's duration, the bridge opens a continuation stream and carries over text history. The team validated voice quality, latency, emotional tone, and interruptions before integration, then expanded from the 13-field form to a conditional intake spanning more than 18 categories. Consent and disclosures precede intake; attorneys retain final decisions. After roughly three months, Stretto reported a 20% relative conversion increase, about 96% field extraction accuracy in testing, approximately eight-minute completed calls, and 95% lower per-minute voice costs than its previous solution. Results were early and not yet broadly rolled out.

### Source excerpt

Stretto replaced its web-form lead capture with Tara, an AI voice agent built on Amazon Nova 2 Sonic that conducts bankruptcy intake interviews by phone and qualifies leads. See how Stretto, the AWS Generative AI Innovation Center, and CI&T designed the voice architecture, validated it for sensitive conversations, and moved it to production.

## SimAI: Mechanistic User Simulation through Psychological Network Models

DevFeed: [SimAI: Mechanistic User Simulation through Psychological Network Models](<https://devfeed.tech/articles/simai-mechanistic-user-simulation-through-psychological-network-models-139910.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21808>)

Author: Cheng Li, Adhyyan Narang, Jiayi Yuan, John F. Buss, Jina Suh, Devon Sandel-Fernandez, Yulia Tsvetkov, Adam M. Kuczynski, Tim Althoff

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [User Experience](<https://devfeed.tech/topics/user-experience.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [alignment](<https://devfeed.tech/tags/alignment.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [diversity](<https://devfeed.tech/tags/diversity.md>), [mental-health](<https://devfeed.tech/tags/mental-health.md>), [network](<https://devfeed.tech/tags/network.md>), [project](<https://devfeed.tech/tags/project.md>), [safety](<https://devfeed.tech/tags/safety.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

SimAI models user responses through interacting psychological factors and trains open-ended simulations with graph-level rewards for reconstructing the underlying relationships. Its PSYCHE dataset contains 18,576 profile-network pairs derived from 516 psychotherapy cases. Automatic evaluations report greater simulation and behavioral diversity than existing patient simulators while maintaining strong profile alignment. In a study with 30 mental-health professionals and trainees, SimAI received the highest average ratings for conversational authenticity, realistic help-seeking, and clinical training utility. The authors also report using it for clinical training and to uncover safety failures in prolonged AI interactions with vulnerable users.

### Source excerpt

arXiv:2611.21808v1 Announce Type: cross Abstract: Realistic user simulation enables the study of human interactions when direct experimentation is costly or potentially harmful. However, existing simulators largely overlook the psychological mechanisms that drive user behavior. We introduce SimAI, a framework for mechanistic user simulation that models how psychological factors interact to shape user responses. To overcome the challenge of supervising open-ended behavior, we propose a graph reconstruction objective that provides graph-level rewards based on how well simulated conversations reflect the underlying psychological relationships. To support this approach, we construct PSYCHE, a dataset of 18,576 profile-network pairs derived from 516 psychotherapy cases. Automatic evaluations show that SimAI achieves greater simulation and behavioral diversity than existing patient simulators while maintaining strong profile alignment. In an interactive study with 30 mental-health professionals and trainees, SimAI receives the highest average ratings for conversational authenticity, realistic help-seeking behavior, and clinical training utility. Beyond simulation quality, SimAI enables realistic clinical training and uncovers safety failures in prolonged AI interactions with vulnerable users. Code is available at https://github.com/behavioral-data/ANGEL-UserSim.

## Conversational analytics with Fivetran & Snowflake Intelligence | Blog | Fivetran

DevFeed: [Conversational analytics with Fivetran & Snowflake Intelligence | Blog | Fivetran](<https://devfeed.tech/articles/conversational-analytics-with-fivetran-snowflake-intelligence-blog-fivetran-140206.md>)

Original publisher: [Read original article](<https://wf.fivetran.com/blog/conversational-analytics-with-fivetran-snowflake-intelligence>)

Author: Kelly Kohlleffel

Published: 2026-11-18T04:00:44Z

Content type: article

Language: en

Sources: [Fivetran Blog](<https://devfeed.tech/sources/fivetran-blog.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [business-intelligence](<https://devfeed.tech/topics/business-intelligence.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [semantic-layer](<https://devfeed.tech/topics/semantic-layer.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-native](<https://devfeed.tech/tags/ai-native.md>), [ai-tools](<https://devfeed.tech/tags/ai-tools.md>), [alerts](<https://devfeed.tech/tags/alerts.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [analyst](<https://devfeed.tech/tags/analyst.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [anomaly-detection](<https://devfeed.tech/tags/anomaly-detection.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [audit](<https://devfeed.tech/tags/audit.md>), [auditability](<https://devfeed.tech/tags/auditability.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [data](<https://devfeed.tech/tags/data.md>), [fivetran](<https://devfeed.tech/tags/fivetran.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [snowflake](<https://devfeed.tech/tags/snowflake.md>)

### AI overview

Snowflake Intelligence lets users ask analytics questions in natural language and can coordinate multi-step workflows across structured and unstructured data. Cortex Analyst queries structured data, Cortex Search retrieves unstructured data, and semantic models connect results to business concepts; the article says governance, observability, and auditability make reasoning and data use visible. Fivetran loads data from varied sources into Snowflake, while Fivetran Activations can route outputs to operational systems. The suggested pilot is to connect a few high-value sources, define semantic views, test an agent against questions with known answers, inspect its reasoning, iterate, then expand sources and agents and set up reports or alerts. The article describes this as an emerging agentic AI use case; it does not provide independent accuracy benchmarks.

### Source excerpt

Fivetran powers the data foundation behind Snowflake's next-generation AI-driven analytics, enabling users to talk directly to their data.

## Where (and when) to deploy conversational AI first

DevFeed: [Where (and when) to deploy conversational AI first](<https://devfeed.tech/articles/where-and-when-to-deploy-conversational-ai-first-142942.md>)

Original publisher: [Read original article](<https://www.twilio.com/en-us/blog/insights/where-to-deploy-conversational-ai-first>)

Author: Jesse Sumrak

Published: 2026-11-18T00:00:00Z

Content type: article

Language: en

Sources: [Twilio](<https://devfeed.tech/sources/twilio-blog.md>), [Twilio Blog](<https://devfeed.tech/sources/twilio-blog-2.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [AI interaction best practices](<https://devfeed.tech/topics/ai-interaction-best-practices.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-deployment](<https://devfeed.tech/tags/ai-deployment.md>), [availability](<https://devfeed.tech/tags/availability.md>), [balance](<https://devfeed.tech/tags/balance.md>), [best](<https://devfeed.tech/tags/best.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [build](<https://devfeed.tech/tags/build.md>), [capture](<https://devfeed.tech/tags/capture.md>), [case-study](<https://devfeed.tech/tags/case-study.md>), [cios](<https://devfeed.tech/tags/cios.md>), [click](<https://devfeed.tech/tags/click.md>), [context](<https://devfeed.tech/tags/context.md>), [conversational-ai](<https://devfeed.tech/tags/conversational-ai.md>), [cost](<https://devfeed.tech/tags/cost.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-journeys](<https://devfeed.tech/tags/customer-journeys.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [industry-insights](<https://devfeed.tech/tags/industry-insights.md>)

### AI overview

Choose a first conversational AI deployment where mistakes are cheap and reversible, the needed data is already accessible in one system, customers face a speed or availability gap, and a staffed human handoff is easy. The article advises against starting with payments, health or signature workflows, emotionally loaded contacts, multi-system journeys, or work that cannot be measured. It recommends capturing handle time, resolution rate, repeat contacts, and abandonment for about two weeks before launch. Examples distinguish WestCX's reversible appointment rescheduling and insurance verification from Eltropy's read-only balance checks and transaction history; Eltropy's containment results are tied to narrowly scoped voice agents at community financial institutions. The text warns that poor escalation can leave customers and agents without context, and that broad, general-purpose deployments are less likely to perform well.

### Source excerpt

Your first conversational AI deployment should be simple and boring. Here's how to choose it, and when you're ready to.

## Nemotron 3.5 Bolt E4B 4GB: An Open and Efficient Small Language Model for Fast Conversational Agents

DevFeed: [Nemotron 3.5 Bolt E4B 4GB: An Open and Efficient Small Language Model for Fast Conversational Agents](<https://devfeed.tech/articles/nemotron-3-5-bolt-e4b-4gb-an-open-and-efficient-small-language-model-for-fast-conversational-agents-137611.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20771>)

Author: NVIDIA, :, Adi Margolin, Aditya Malte, Aishwarya Chhabra, Alex Hodson, Ameya Sunil Mahabaleshwarkar, Ambrish Dantrey, Amy Shen, Andrii Skliar, Annamalai Chockalingam, Anusha Pant, Ashwath Aithal, Ashw

Published: 2026-11-17T05:00:00Z

Content type: release

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Nemotron](<https://devfeed.tech/topics/nemotron.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Local AI](<https://devfeed.tech/topics/local-ai.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [concurrency](<https://devfeed.tech/tags/concurrency.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decode](<https://devfeed.tech/tags/decode.md>), [dflash](<https://devfeed.tech/tags/dflash.md>), [dgx-spark](<https://devfeed.tech/tags/dgx-spark.md>), [edge](<https://devfeed.tech/tags/edge.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [fast](<https://devfeed.tech/tags/fast.md>), [geforce](<https://devfeed.tech/tags/geforce.md>), [geforce-rtx](<https://devfeed.tech/tags/geforce-rtx.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [nemotron](<https://devfeed.tech/tags/nemotron.md>), [nemotron-3-5](<https://devfeed.tech/tags/nemotron-3-5.md>), [nemotron-3-5-bolt](<https://devfeed.tech/tags/nemotron-3-5-bolt.md>), [open](<https://devfeed.tech/tags/open.md>), [small](<https://devfeed.tech/tags/small.md>), [small-language-model](<https://devfeed.tech/tags/small-language-model.md>)

### AI overview

Nemotron 3.5 Bolt E4B 4GB is an open small language model post-trained from Gemma-4-E4B-it for fast conversational agents, with reported improvements in multi-turn tool use, instruction following, and input grounding. Across 21 benchmarks, the authors report the highest average accuracy and fewest output tokens in reasoning and non-reasoning modes versus Gemma-4-E4B-it and Qwen3.5-4B. The release includes a DFlash drafter: at concurrency 1, it reaches 2.74x the per-user decode speed of inference without speculative decoding; at batch size 1, it achieves 1.21x Gemma-4-E4B-it's generation throughput with MTP. These figures reflect the configurations stated in the abstract.

### Source excerpt

arXiv:2611.20771v1 Announce Type: cross Abstract: We introduce Nemotron 3.5 Bolt E4B 4GB, an open and efficient small language model further post-trained from Gemma-4-E4B-it and optimized for speed and intelligence per GB of memory used for fast conversational agents. The model carries forward Gemma's strong conversational capabilities and further improves multi-turn tool use, instruction following, and input grounding for fast, local voice/device assistants, interactive personas, and embedded/Edge interfaces. The model combines natural conversation with multi-turn agentic capabilities to power voice/device assistants, interactive characters and robotic/IoT interfaces running on GeForce RTX, Jetson, and RTX/DGX Spark. Experiments on 21 benchmarks show that Nemotron 3.5 Bolt E4B 4GB achieves the highest average accuracy while generating the fewest output tokens in both reasoning and non-reasoning modes, compared with Gemma-4-E4B-it and Qwen3.5-4B. The model is released with a DFlash drafter. With DFlash, Bolt E4B achieves $2.74\times$ the per-user decode speed of inference without speculative decoding at concurrency~1, and $1.21\times$ the generation throughput of Gemma-4-E4B-it with MTP at batch size 1.

## Conversing with a Conversational Agent: A Real Conversation, Not a Simulated One

DevFeed: [Conversing with a Conversational Agent: A Real Conversation, Not a Simulated One](<https://devfeed.tech/articles/conversing-with-a-conversational-agent-a-real-conversation-not-a-simulated-one-137618.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20826>)

Author: Christophe Denis (UMMISCO, IHPST)

Published: 2026-11-17T05:00:00Z

Content type: opinion

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [consciousness](<https://devfeed.tech/tags/consciousness.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [llm](<https://devfeed.tech/tags/llm.md>), [meaning](<https://devfeed.tech/tags/meaning.md>), [mirror](<https://devfeed.tech/tags/mirror.md>), [philosophy](<https://devfeed.tech/tags/philosophy.md>), [process](<https://devfeed.tech/tags/process.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [systems](<https://devfeed.tech/tags/systems.md>)

### AI overview

The article argues that exchanges with LLM-based agents are real conversational activities, even though the agents need not be conscious, subjectively understand, or possess human intentionality. It challenges definitions of conversation based only on participants' intrinsic properties and develops a relational, semiotic account. Its concept of "Hybrid Semiotic Fabrication" describes meaning-making as distributed across people, AI, corpora, and technical arrangements. The argument separates the reality of interaction from apparent artificial subjectivity and from whether produced meanings are epistemically valid: treating a conversation as real does not make the agent human or its outputs authoritative.

### Source excerpt

arXiv:2611.20826v1 Announce Type: new Abstract: The claim that interactions with artificial conversational agents merely simulate human conversation rests on a philosophical distinction between reproducing the observable features of an activity and actually performing it. Although this distinction is appropriate for certain physical phenomena, its application to linguistic and symbolic activities is far less straightforward. This article argues that conversations with large language model (LLM)-based agents are genuine conversational activities rather than mere simulations, without thereby attributing consciousness, subjective understanding, or human intentionality to artificial systems. Drawing on the distinction between subjective and objective simulation, the article first examines the limitations of defining conversation exclusively through the intrinsic properties of its participants. It then develops a relational and semiotic account of conversational activity, informed by linguistic pragmatics, Lacan's Conversing with a Conversational Agent theory of the mirror, Baudrillard's concept of the simulacrum, Simondon's philosophy of individuation, and Peirce's semiotics. These perspectives make it possible to distinguish the reality of conversational interaction from the appearance of artificial subjectivity and from the epistemic validity of the meanings produced. The argument culminates in the concept of Hybrid Semiotic Fabrication, which characterizes meaning-making as a distributed process involving human participants, artificial intelligence, corpora, and technical arrangements. Recognizing the reality of human-AI conversation therefore entails neither anthropomorphizing artificial agents nor granting their outputs epistemic authority. Rather, it requires a more precise account of the relational, semiotic, and normative conditions under which conversational meaning is produced and justified.

## FRAAC: A French Multimodal Large Language Model and Dataset forPictogram-Based Augmentative and Alternative Communication

DevFeed: [FRAAC: A French Multimodal Large Language Model and Dataset forPictogram-Based Augmentative and Alternative Communication](<https://devfeed.tech/articles/fraac-a-french-multimodal-large-language-model-and-dataset-forpictogram-based-augmentative-and-alternative-communication-137533.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20094>)

Author: Alexandre Audibert, Fran\\c{c}ois Portet, Benoit Favre, Maja Jonck Hjuler

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>)

Tags: [annotation](<https://devfeed.tech/tags/annotation.md>), [communication](<https://devfeed.tech/tags/communication.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [language](<https://devfeed.tech/tags/language.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [pairs](<https://devfeed.tech/tags/pairs.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [rank](<https://devfeed.tech/tags/rank.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [rules](<https://devfeed.tech/tags/rules.md>), [sequences](<https://devfeed.tech/tags/sequences.md>), [speech](<https://devfeed.tech/tags/speech.md>), [text](<https://devfeed.tech/tags/text.md>)

### AI overview

FRAAC addresses the lack of parallel French text-pictogram data and reliable pictogram handling in multimodal language models. Its Retrieval-Augmented Pictogram Selection (RAPS) method refines rule-based transcriptions by extracting concepts with an LLM and retrieving pictograms, covering named entities, numbers, and polysemous words without adding hand-written rules. Applied to French, it produced 1.4 million synthetic parallel pairs, with 300,000 publicly released. In a human ranking study, RAPS outperformed the rule-based state of the art. The authors report that reference-based metrics distinguish clearly different quality levels but do not reliably rank strong systems. FRAAC also includes an instruction-tuned model that accepts text or pictogram input and responds with pictograms, speech, or text.

### Source excerpt

arXiv:2611.20094v1 Announce Type: cross Abstract: Pictograms are a core modality in augmentative and alternative communication (AAC), yet recent multimodal large language models (MLLMs) cannot reliably interpret or produce pictogram sequences. Two obstacles explain this gap: the absence of a standardized pictogram grammar and the scarcity of parallel text-pictogram data. We introduce Retrieval-Augmented Pictogram Selection (RAPS), an annotation method that refines a rule-based transcription system's output with LLM-based concept extraction and retrieval over a pictogram vocabulary, handling cases where hand-written rules fail (named entities, numbers, polysemous words) without requiring new rules. Applied to French, RAPS yields a synthetic corpus of 1.4 million parallel pairs, of which 300k are publicly released. In a human ranking study where annotators order four pictogram transcriptions of the same sentence, RAPS outperforms the rule-based state of the art. Moreover, we show that standard reference-based metrics separate systems of clearly different quality but fail to rank strong systems against each other. Finally, we build an instruction-tuned model that supports interaction in several AAC modalities: it accepts pictogram or text input and answers in pictograms, speech, or text.

## Does the Voice Matter? Evaluating Frustration Detection on Deployed Clinical Voice-Agent Calls

DevFeed: [Does the Voice Matter? Evaluating Frustration Detection on Deployed Clinical Voice-Agent Calls](<https://devfeed.tech/articles/does-the-voice-matter-evaluating-frustration-detection-on-deployed-clinical-voice-agent-calls-137644.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21001>)

Author: Annika Michael, Zachary Ellis, Yuxiang Zhou, Samuel Thio, Yajie Vera He, Spencer Hazel, Jared Joselowitz

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [asr](<https://devfeed.tech/topics/asr.md>), [speech-to-speech](<https://devfeed.tech/topics/speech-to-speech.md>)

Tags: [audio](<https://devfeed.tech/tags/audio.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [detection](<https://devfeed.tech/tags/detection.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [speech-to-text](<https://devfeed.tech/tags/speech-to-text.md>), [transcript](<https://devfeed.tech/tags/transcript.md>), [triage](<https://devfeed.tech/tags/triage.md>), [voice](<https://devfeed.tech/tags/voice.md>), [voice-agents](<https://devfeed.tech/tags/voice-agents.md>)

### AI overview

In 30 deployed clinical voice-agent calls, patient turns were often too short for transcripts alone to reliably detect frustration: about 59% were one or two words, and transcript-based detection flagged only 3 or 4 of 39 frustrated one-word turns. Adding audio did not consistently improve results. Two Gemini 2.5 models gained 3.0 percentage points when prompted to attend to vocal delivery, but no model showed a statistically clear audio advantage when given the same prompt for both inputs; some models performed worse with audio under input-specific prompts. A WavLM probe reached 71.6% overall accuracy yet missed every frustrated one-word turn in 24 added calls. The findings indicate that a short transcript cannot rule out frustration, while audio benefits depend on the model and prompt.

### Source excerpt

arXiv:2611.21001v1 Announce Type: cross Abstract: Autonomous voice agents now conduct routine clinical follow-up calls, so some automated triage is unavoidable. A patient becoming frustrated is one thing that triage should catch. These systems typically read only the transcript the speech-to-text stage produces, not the voice, and whether that is enough is largely untested on real clinical speech. We study per-turn frustration detection on real patient calls from a deployed clinical voice agent: 150 turns from six calls labelled by three annotators, with a length-matched acted corpus as a control, and 714 turns from 30 calls. Patient turns are overwhelmingly short: across the deployed calls, roughly 59% are one or two words, against 13% in the full acted corpus, and a corpus accuracy gap remains even after the acted set is length-matched. On one-word turns, transcript-only detection does not exclude chance. Across five audio-capable language models, the transcript flags 3 or 4 of the 39 frustrated one-word turns in the 30 calls, and a classifier given only the word count reaches 57.3%. Whether audio helps depends on the model and on the instruction. Asked to attend to delivery, two Gemini 2.5 models gain 3.0 pp over the transcript across 30 calls (-1.0 to 7.5 pp). Of the turns they confidently clear, 6.9% were in fact frustrated for audio against 13.0% for the transcript, a difference of 6.1 pp (2.2 to 10.4 pp). Across the 30 calls, given one prompt identical for both inputs, no model's audio-text interval excludes zero. Under the input-specific prompts, a newer Gemini model and two smaller open models do worse with audio than with the transcript (-3.5, -8.6 and -10.6 pp). A WavLM probe trained on the clinical calls matches the best prompted model overall (71.6%) but misses every frustrated one-word turn in the 24 added calls. When a patient's turn is short, the transcript alone is not enough to rule out frustration.

## Share your Screen and Talk to Gemini: How Young Learners Interact with Multimodal GenAI While Solving Scratch Tasks

DevFeed: [Share your Screen and Talk to Gemini: How Young Learners Interact with Multimodal GenAI While Solving Scratch Tasks](<https://devfeed.tech/articles/share-your-screen-and-talk-to-gemini-how-young-learners-interact-with-multimodal-genai-while-solving-scratch-tasks-137680.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21306>)

Author: Sven Jacobs, Lukas Berning, Natalie Kiesler

Published: 2026-11-17T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [human computer interaction](<https://devfeed.tech/topics/human-computer-interaction.md>), [AI response quality](<https://devfeed.tech/topics/ai-response-quality.md>), [Learning](<https://devfeed.tech/topics/learning.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cy](<https://devfeed.tech/tags/cs-cy.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [gemini](<https://devfeed.tech/tags/gemini.md>), [genai](<https://devfeed.tech/tags/genai.md>), [learn-programming](<https://devfeed.tech/tags/learn-programming.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [screen](<https://devfeed.tech/tags/screen.md>), [tutor](<https://devfeed.tech/tags/tutor.md>)

### AI overview

After six Scratch lessons, 23 fifth- and sixth-graders used a screen-sharing GenAI tutor for eight tasks. Its replies were intelligible and actionable, and no single reply disclosed a complete solution. But it missed existing errors in 61.5% of 603 error-related classroom turns, versus 15.7% in a pilot conducted without students. Children's voice prompts were brief and mostly responses to the tutor; despite being told answers could be wrong, 39.1% trusted them to always be correct. The findings suggest that a human-like voice does not guarantee accurate help, and screen sharing alone may not align responses with the program's state.

### Source excerpt

arXiv:2611.21306v1 Announce Type: new Abstract: Children using Scratch to learn programming can struggle with text-heavy, lengthy explanations and hints during debugging. Verbal interactions based on screen sharing or camera data may be an interesting alternative for supporting learners. We therefore explore the quality, use, and perceptions of fifth- and sixth-graders' verbal interactions with a GenAI tutor. The goal is to investigate the quality and use of the multimodal GenAI tutor in the given context. First of all, a pilot study of eight tasks conducted without students helped inform the process of the tutor's classroom evaluation. The classroom study was realized after fifth- and sixth-graders had participated in six Scratch classroom lessons. Then 23 students were introduced to the tutor, worked on the eight tasks with the GenAI tutor, and completed a questionnaire. The analysis of the pilot and classroom study shows that the tutor's responses were intelligible and actionable, and no single turn revealed a complete solution. However, the tutor missed existing errors in 61.5% of 603 error-related classroom turns, compared to 15.7% in the pilot study. Children's voice prompts were brief and mostly responded to the tutor. Although a video had explained that responses could be incorrect, 39.1% of the children trusted the tutor's answers to always be correct. These findings have implications for educators and learners using multimodal GenAI tutors, as they need to be aware that an intelligible and human-like voice does not ensure correct support. For developers, our results suggest that screen sharing alone does not ensure responses that match the program state.

## MemoryLane: Supporting Reminiscence Conversations Between Older Adults with Mild Dementia and Lay Caregivers Using a Simulated AI Agent

DevFeed: [MemoryLane: Supporting Reminiscence Conversations Between Older Adults with Mild Dementia and Lay Caregivers Using a Simulated AI Agent](<https://devfeed.tech/articles/memorylane-supporting-reminiscence-conversations-between-older-adults-with-mild-dementia-and-lay-caregivers-using-a-simulated-ai-agent-137376.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19013>)

Author: Soumay Agarwal, Aaryan Gajula, Lingyao Li, Renkai Ma, Jennifer Martin, Krisstina Madan, Ellen Brown, Chen Chen

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [human computer interaction](<https://devfeed.tech/topics/human-computer-interaction.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Image](<https://devfeed.tech/topics/image.md>), [human review](<https://devfeed.tech/topics/human-review.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-generated-images](<https://devfeed.tech/tags/ai-generated-images.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [study](<https://devfeed.tech/tags/study.md>)

### AI overview

MemoryLane helps lay caregivers facilitate reminiscence conversations without curating personal photographs. It brings the older adult, caregiver and a simulated AI peer into a triadic conversation, showing the agent and AI-generated images of discussed memories on a shared interface. A private caregiver smartphone interface provides controls for the agent's behavior and moderation of generated images. A preliminary focus group with five dementia-care stakeholders suggested the peer companion and generated images could be valuable, and that caregiver controls and image moderation were feasible. The evidence provided is an abstract and does not establish clinical outcomes.

### Source excerpt

arXiv:2611.19013v1 Announce Type: cross Abstract: Reminiscence Therapy (RT) is a non-pharmacological intervention that invites patients to discuss and review past life experiences and memories to improve their emotional and mental well-being. In practice, many reminiscence conversations are led by lay caregivers, such as spouses and adult children. While traditional RT often relies on personal photographs, curating them can be tedious and difficult. In addition, caregivers who lack shared life experiences with the patient may struggle to facilitate engaging reminiscence conversations. This paper presents MemoryLane, an AI-integrated tool that helps lay caregivers facilitate reminiscence conversations without curated photographs. MemoryLane supports a triadic conversation among the patient, the caregiver, and a simulated AI agent that acts as the patient's peer. A shared interface shows the agent and AI-generated images of the memories being discussed, while a private smartphone interface lets the caregiver control the behavior of the simulated AI agent and moderate the AI-generated images. A preliminary focus group study with five dementia-care stakeholders suggested the value of MemoryLane's AI agent as a peer companion and of AI-generated images, as well as the feasibility of having caregivers moderate images and control the AI agent.

## From Voice AI Demo to Production: Why Infrastructure and Agent Supervision Matter

DevFeed: [From Voice AI Demo to Production: Why Infrastructure and Agent Supervision Matter](<https://devfeed.tech/articles/from-voice-ai-demo-to-production-why-infrastructure-and-agent-supervision-matter-140356.md>)

Original publisher: [Read original article](<https://www.twilio.com/en-us/blog/products/new-capabilities-conversation-relay-intelligence-twilio-agent-connect>)

Author: Lenore Files

Published: 2026-11-17T00:00:00Z

Content type: release

Language: en

Sources: [Twilio](<https://devfeed.tech/sources/twilio-blog.md>), [Twilio Blog](<https://devfeed.tech/sources/twilio-blog-2.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [agent observability](<https://devfeed.tech/topics/agent-observability.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [observability](<https://devfeed.tech/topics/observability.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-behavior](<https://devfeed.tech/tags/agent-behavior.md>), [agent-frameworks](<https://devfeed.tech/tags/agent-frameworks.md>), [agent-runtime](<https://devfeed.tech/tags/agent-runtime.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-powered](<https://devfeed.tech/tags/ai-powered.md>), [ai-stack](<https://devfeed.tech/tags/ai-stack.md>), [connect](<https://devfeed.tech/tags/connect.md>), [conversation-intelligence](<https://devfeed.tech/tags/conversation-intelligence.md>), [conversation-relay](<https://devfeed.tech/tags/conversation-relay.md>), [demo](<https://devfeed.tech/tags/demo.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [product-news](<https://devfeed.tech/tags/product-news.md>), [production](<https://devfeed.tech/tags/production.md>), [supervision](<https://devfeed.tech/tags/supervision.md>), [voice](<https://devfeed.tech/tags/voice.md>), [voice-ai](<https://devfeed.tech/tags/voice-ai.md>)

### AI overview

Twilio presents production Voice AI as a combination of real-time communications reliability, flexibility across AI providers, and supervision of actual customer conversations. Conversation Relay adds Smart Turn Detection to reduce premature replies, Automated STT Failover to switch to a backup provider during latency, rate limits, or stream disruptions, mid-call updates to speech hints, language, and voice settings, HD Audio, and support for ElevenLabs v4 Turbo. Twilio Agent Connect connects external agent frameworks and LLM runtimes to Twilio voice and messaging, letting teams keep their chosen models and business logic while Twilio supplies the communications layer. Conversation Intelligence offers transcript-first review without agent-code changes or intrusive SDKs; its public-beta Twilio-authored operators assess signals such as task completion and hallucinations, while its GA On-Demand Execution API runs Language Operators on demand alongside automated triggers. The article positions these features as tools for continuity, provider choice, and quality monitoring at scale, not as Twilio-provided agents or models.

### Source excerpt

Explore Twilio's new Voice AI infrastructure features from Conversation Relay, Agent Connect, and Conversation Intelligence.

## Jordan Belfort's new AI voice startup, Tone, wants to sell like the 'Wolf of Wall Street'

DevFeed: [Jordan Belfort's new AI voice startup, Tone, wants to sell like the 'Wolf of Wall Street'](<https://devfeed.tech/articles/jordan-belfort-s-new-ai-voice-startup-tone-wants-to-sell-like-the-wolf-of-wall-street-136060.md>)

Original publisher: [Read original article](<https://techcrunch.com/2026/11/16/jordan-belforts-new-ai-voice-startup-tone-wants-to-sell-like-the-wolf-of-wall-street/>)

Author: Connie Loizos

Published: 2026-11-16T17:10:00Z

Content type: news

Language: en

Sources: [TechCrunch | AI News & Artificial Intelligence](<https://devfeed.tech/sources/techcrunch-ai-news-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Chat Bot](<https://devfeed.tech/topics/chatbot.md>)

Tags: [22](<https://devfeed.tech/tags/22.md>), [about](<https://devfeed.tech/tags/about.md>), [about-you](<https://devfeed.tech/tags/about-you.md>), [act](<https://devfeed.tech/tags/act.md>), [added](<https://devfeed.tech/tags/added.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-voice](<https://devfeed.tech/tags/ai-voice.md>), [best](<https://devfeed.tech/tags/best.md>), [can](<https://devfeed.tech/tags/can.md>), [deals](<https://devfeed.tech/tags/deals.md>), [jordan](<https://devfeed.tech/tags/jordan.md>), [like](<https://devfeed.tech/tags/like.md>), [sales](<https://devfeed.tech/tags/sales.md>), [salespeople](<https://devfeed.tech/tags/salespeople.md>), [startup](<https://devfeed.tech/tags/startup.md>), [tc](<https://devfeed.tech/tags/tc.md>), [up](<https://devfeed.tech/tags/up.md>), [voice](<https://devfeed.tech/tags/voice.md>), [wall-street](<https://devfeed.tech/tags/wall-street.md>), [wants](<https://devfeed.tech/tags/wants.md>), [well](<https://devfeed.tech/tags/well.md>)

### AI overview

Tone is a voice AI sales platform built around Jordan Belfort's claim that strategically chosen vocal tones can improve persuasion. He says it handles full sales conversations, including qualifying customers, explaining insurance plans, adjusting options, and returning to a planned flow after questions; reported latency is about 600 milliseconds. Its customers include law firms, health insurers, polling and automobile businesses, which Belfort says use it when leads exceed staff capacity or they need coverage across hours or languages. He says Tone uses small, fast models, and concedes it does not yet match a top human salesperson, though he expects that to change; he predicts it can outperform a mediocre closer. Belfort says calls are limited to opted-in leads or existing customers, with polling cited as an exception, and that voice cloning requires consent and identity checks. Tone charges per minute for streaming, monthly for its planned content narration platform, and through success-based arrangements for some clients. The article says it could not independently verify the technology's performance.

### Source excerpt

Belfort claims his AI sales voice can already close deals almost as well as the best human salespeople, and he thinks it will catch up in a year.

## SA-RAG: Structuring Noisy Conversations for Retrieval-Augmented Generation

DevFeed: [SA-RAG: Structuring Noisy Conversations for Retrieval-Augmented Generation](<https://devfeed.tech/articles/sa-rag-structuring-noisy-conversations-for-retrieval-augmented-generation-135692.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.24366>)

Author: Kaiqiao Han, LuAn Tang, Renliang Sun, Peng Yuan, Wei Cheng, Haoyu Wang, Wei Wang, Yizhou Sun, Haifeng Chen

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>), [Enhancing AI model performance](<https://devfeed.tech/topics/enhancing-ai-model-performance.md>), [AI-driven information retrieval](<https://devfeed.tech/topics/ai-driven-information-retrieval.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>)

Tags: [consistency](<https://devfeed.tech/tags/consistency.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [external](<https://devfeed.tech/tags/external.md>), [framework](<https://devfeed.tech/tags/framework.md>), [induction](<https://devfeed.tech/tags/induction.md>), [information-retrieval](<https://devfeed.tech/tags/information-retrieval.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llms](<https://devfeed.tech/tags/llms.md>), [quality](<https://devfeed.tech/tags/quality.md>), [rag](<https://devfeed.tech/tags/rag.md>), [rag-applications](<https://devfeed.tech/tags/rag-applications.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [reliability](<https://devfeed.tech/tags/reliability.md>), [retrieval-augmented-generation](<https://devfeed.tech/tags/retrieval-augmented-generation.md>), [robust](<https://devfeed.tech/tags/robust.md>), [semantic](<https://devfeed.tech/tags/semantic.md>)

### AI overview

SA-RAG aims to make retrieval-augmented generation more robust to noisy conversational data by inducing intermediate table structures directly from it, rather than assuming clean, high-quality structures are already available. Its quality-aware induction jointly models semantic and structural consistency. The paper explores both training-free and training-based strategies to improve structure quality while preserving fidelity. Experiments on two noisy real-world datasets reportedly show significant gains over existing RAG baselines; further analysis emphasizes the importance of high-quality structures. The supplied text does not specify the datasets, metrics, or strategy details.

### Source excerpt

arXiv:2605.24366v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have been widely adopted in conversational applications. However, their reliance on parametric knowledge limits reliability in real-world scenarios that require dynamic or domain-specific information. Retrieval-Augmented Generation (RAG) addresses this limitation by incorporating external knowledge during generation, but existing RAG methods often struggle with noisy or irrelevant contexts, as they typically assume that high-quality structures are already available. In this work, we propose Structure-aware Retrieval-Augmented Generation (SA-RAG), a principled structure induction and generation framework that constructs reliable intermediate structures directly from noisy conversational data. SA-RAG uses tables as compact and controllable structured representations. We introduce a quality-aware structure induction framework that jointly models semantic and structural consistency. Furthermore, we explore both training-free and training-based structure generation strategies to further improve structure quality while maintaining fidelity. Experiments on two noisy real-world datasets show that SA-RAG significantly outperforms existing RAG baselines. Further analysis highlights that high-quality structural representations are crucial for robust RAG applications.

## A frontier machine intelligence for human neuroimaging

DevFeed: [A frontier machine intelligence for human neuroimaging](<https://devfeed.tech/articles/a-frontier-machine-intelligence-for-human-neuroimaging-132093.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.15094>)

Author: Chenhui Zhao, Akshay Rao, Akhil Kondepudi, Yiwei Lyu, Xinhai Hou, Rikhil Seshadri, Cheng Jiang, Soumyanil Banerjee, Rushikesh Joshi, Samir Harake, Ashok Srinivasan, Shruti Mishra, Francisco Rivas-Rodr

Published: 2026-11-13T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Medical imaging](<https://devfeed.tech/topics/medical-imaging.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [ai evaluation](<https://devfeed.tech/topics/ai-evaluation.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [frontier-models](<https://devfeed.tech/tags/frontier-models.md>), [interactive](<https://devfeed.tech/tags/interactive.md>), [learning](<https://devfeed.tech/tags/learning.md>), [medical-ai](<https://devfeed.tech/tags/medical-ai.md>), [medical-imaging](<https://devfeed.tech/tags/medical-imaging.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [reports](<https://devfeed.tech/tags/reports.md>), [vision-language-model](<https://devfeed.tech/tags/vision-language-model.md>)

### AI overview

Neuro-VL is a vision-language model trained on nearly one million neuroimaging studies from routine clinical care. The abstract reports that it interprets complete examinations, discusses findings in natural language, generates radiology reports, and compares serial scans to assess disease progression. Across internal and international datasets, it reportedly outperformed general-purpose and specialized medical models on diagnosis, abnormality localization and description, and report generation. It also more reliably resisted misleading clinical assumptions than leading models from OpenAI and Google. These are reported study results; the supplied text does not provide performance metrics or further methodological detail.

### Source excerpt

arXiv:2611.15094v1 Announce Type: new Abstract: Medical imaging is essential for diagnosing disease, guiding treatment and monitoring patients over time. Although artificial intelligence has advanced image interpretation, existing medical models often address restricted tasks, while general-purpose proprietary models lack access to private clinical imaging required to achieve medical expertise. Developing models that can interpret complex examinations and communicate their findings in the context of clinical decisions remains a major challenge. Here, we show that Neuro-VL, a vision-language model trained on nearly one million neuroimaging studies from routine clinical care, can interpret complete imaging examinations and engage in natural language conversations about their findings. Across internal and international datasets, Neuro-VL outperformed frontier general-purpose and specialized medical models in diagnosing, localizing, and describing abnormalities, as well as generating comprehensive radiology reports. The model also compared serial examinations to assess disease progression and more reliably resisted misleading clinical assumptions than leading frontier models from OpenAI and Google. These results extend health system learning beyond individual predictions to interactive image interpretation and longitudinal disease monitoring. They demonstrate how the collective clinical experience of health systems can be used to develop medical AI with specialized expertise and broad clinical capabilities.

## Kezek: Splitting Mono Two-Person Conversations into Per-Speaker Tracks for Full-Duplex Dialogue Data

DevFeed: [Kezek: Splitting Mono Two-Person Conversations into Per-Speaker Tracks for Full-Duplex Dialogue Data](<https://devfeed.tech/articles/kezek-splitting-mono-two-person-conversations-into-per-speaker-tracks-for-full-duplex-dialogue-data-132293.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16933>)

Author: Bagzhan Karl, Alizhan Karl

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Open Source Models & Datasets](<https://devfeed.tech/topics/open-source-models-datasets.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [audio](<https://devfeed.tech/tags/audio.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [eess-as](<https://devfeed.tech/tags/eess-as.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [mono](<https://devfeed.tech/tags/mono.md>), [replies](<https://devfeed.tech/tags/replies.md>), [separator](<https://devfeed.tech/tags/separator.md>), [speaker](<https://devfeed.tech/tags/speaker.md>)

### AI overview

Kezek converts mono recordings of exactly two people into two full-length, time-aligned speaker tracks, preserving overlaps and short replies on the correct speaker's track. It first determines speaker identity and activity for the whole recording, then uses a conditioned time-domain separator to implement that assignment; the two outputs sum to the input. Compared with pyannote followed by DialogueSidon, Kezek retains more short replies overall on Open Yap and CallHome, largely because it covers the full recording. On audio both pipelines return, reply retention is similar (0.859 vs. 0.865 on Open Yap). The abstract also reports less leakage and fewer channel swaps, and 4-5x faster processing, at about 230 audio hours per GPU-hour. Quiet replies masked by a louder voice remain a challenge. These findings are from the supplied abstract.

### Source excerpt

arXiv:2611.16933v1 Announce Type: cross Abstract: Full-duplex spoken dialogue models learn turn-taking, overlaps and backchannels from conversations recorded with one channel per speaker, but most available conversational audio is a single mono mix. We present Kezek, a model that turns a mono recording of exactly two people into one track per person with the original length and timing; overlaps and short replies ("yeah", "mm-hm") stay on the track of the person who said them. Kezek decides who speaks when and who is who once for the whole recording (stage 1: a two-voice activity model guided by a speaker fingerprint trained to be insensitive to phase), and a conditioned time-domain separator only carries out that decision (stage 2), with outputs that always sum to the input. Standard separation metrics miss the failures that matter for dialogue data, so we report short-reply retention, leakage onto the other person's solo, channel swaps and coverage. Against the open pipeline pyannote -> DialogueSidon on the same recordings, Kezek keeps more short replies (0.83 vs. 0.69 on Open Yap test; 0.80 vs. 0.71 over the other's speech on unseen telephone speech from CallHome English), mostly because it returns the whole recording while the external pipeline drops part of it (on the audio both return, reply retention is about the same: 0.859 vs. 0.865 on Open Yap); it also leaks less, swaps channels less often, and runs 4-5 times faster (about 230 hours of audio per GPU-hour). We also report negative results and an ideal-mask ceiling that locate the remaining gap: quiet replies under a louder voice. Code and weights are released under permissive licences.

## Scalable Conversational Assessment: Chat-Based Exams with a Two-Agent LLM Pipeline

DevFeed: [Scalable Conversational Assessment: Chat-Based Exams with a Two-Agent LLM Pipeline](<https://devfeed.tech/articles/scalable-conversational-assessment-chat-based-exams-with-a-two-agent-llm-pipeline-132322.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17220>)

Author: Geoffrey Challen, Craig Zilles

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>)

Tags: [adversarial](<https://devfeed.tech/tags/adversarial.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-architecture](<https://devfeed.tech/tags/agent-architecture.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-literacy](<https://devfeed.tech/tags/ai-literacy.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [argument](<https://devfeed.tech/tags/argument.md>), [assessment](<https://devfeed.tech/tags/assessment.md>), [chat](<https://devfeed.tech/tags/chat.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [course](<https://devfeed.tech/tags/course.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cy](<https://devfeed.tech/tags/cs-cy.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [exams](<https://devfeed.tech/tags/exams.md>), [generation](<https://devfeed.tech/tags/generation.md>), [guidance](<https://devfeed.tech/tags/guidance.md>), [interactive](<https://devfeed.tech/tags/interactive.md>), [knowledge](<https://devfeed.tech/tags/knowledge.md>), [llm](<https://devfeed.tech/tags/llm.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [prompt](<https://devfeed.tech/tags/prompt.md>)

### AI overview

The system uses two LLM agents to scale interactive, text-based assessment of students' arguments. An interviewer talks with the student and receives only guidance on the next question; a hidden evaluator tracks rubric progress and directs whether to follow up or move on, reducing the chance of revealing answers. The authors tested it with simulated student personas, including adversarial tactics, using a fictional topic to limit reliance on pretrained knowledge, and deployed it in formative and summative settings in a small elective AI literacy course. Transcript reviews identified needs to improve follow-up questions and ensure grading credits student contributions rather than ideas supplied by the interviewer.

### Source excerpt

arXiv:2611.17220v1 Announce Type: cross Abstract: Oral exams are compelling because (1) students can be asked to provide their reasoning for an answer and (2) their interactive nature permits follow-up questions to resolve ambiguous answers. But oral exams scale poorly. In this paper, we present scalable conversational assessment, an LLM-based system that conducts interactive text-based conversations with students to assess their ability to make and support an argument. We use a two-agent architecture to separate concerns and reduce the likelihood that answers are leaked to the student: the interviewer interacts directly with the student, but only receives guidance about the next question to ask the student; the hidden evaluator knows the goals of the whole activity, tracks the student's progress using a rubric, and guides the interviewer on when to follow-up on the current question or proceed to the next question. The system was first tested using a collection of simulated students with LLM personas of a particular ability level or adversarial tactic (e.g., answer-extractors, prompt-injectors, confident fabricators). A novel element of this evaluation was that it required the use of a fictional assessment topic to prevent personas that shouldn't have information from using knowledge present in the LLM's pre-training. The system was also deployed in a small, elective AI literacy course in both formative and summative contexts. Through manual and post-facto AI review of the transcripts we identified opportunities to improve the follow-up question generation and ensure that grading only credited ideas that the student (and not the interviewer) contributed.

## AURA 2.0: From User-Driven Responses to Model-Initiated Conversations in Streaming Video

DevFeed: [AURA 2.0: From User-Driven Responses to Model-Initiated Conversations in Streaming Video](<https://devfeed.tech/articles/aura-2-0-from-user-driven-responses-to-model-initiated-conversations-in-streaming-video-132306.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17048>)

Author: Shuhan Li, Huankang Guan, Xintong Guo, Yang Bo, Yuyang He, Dunyuan Xu, Jingyan Chen, Fang Liu, Tsz Kin Wan, Jinpeng Chen, Zhenqi He, Yuxuan Hu, Wenbo Li, Han Gao, Qijun Li, Long Chen, Hongsheng Li, Ru

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [vlm](<https://devfeed.tech/topics/vlm.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Real-time AI interactions](<https://devfeed.tech/topics/real-time-ai-interactions.md>), [function calling](<https://devfeed.tech/topics/function-calling.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [ai-assistants](<https://devfeed.tech/tags/ai-assistants.md>), [aura](<https://devfeed.tech/tags/aura.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [grounding](<https://devfeed.tech/tags/grounding.md>), [information](<https://devfeed.tech/tags/information.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [interactive](<https://devfeed.tech/tags/interactive.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [memory](<https://devfeed.tech/tags/memory.md>), [model-weights](<https://devfeed.tech/tags/model-weights.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [performance](<https://devfeed.tech/tags/performance.md>), [streaming](<https://devfeed.tech/tags/streaming.md>), [streaming-video](<https://devfeed.tech/tags/streaming-video.md>), [tool-calling](<https://devfeed.tech/tags/tool-calling.md>), [tools](<https://devfeed.tech/tags/tools.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

AURA 2.0 extends streaming video assistants beyond user-prompted replies: Model-Initiated Conversation (MIC) lets it start discussions about events in an evolving video, while Stream-Aware Tool Calling (SATC) uses visual and conversational context to call external tools and incorporate their results. Interactive Visual Grounding (IVG) handles user-marked objects and returns frame-aligned bounding boxes linking responses to regions. The system is trained on a dedicated streaming interaction dataset using a three-stage scheme, and Long-Horizon Interaction Memory (LHIM) supports continuity across extended streams. The authors report leading performance among open-source streaming models on public streaming benchmarks and positive targeted evaluations of the new capabilities; the abstract provides no comparative scores or further evaluation details. Weights and inference code are released.

### Source excerpt

arXiv:2611.17048v1 Announce Type: cross Abstract: Most AI assistants respond only when prompted, remaining silent even when events in the video stream offer meaningful opportunities for conversation. To move beyond purely user-driven interaction, we present AURA 2.0, a streaming video large language model that introduces three core upgrades to AURA. The key advance is Model-Initiated Conversation (MIC), which enables the model to identify topics worth discussing in the evolving visual stream and initiate conversations at appropriate moments without a user query. Stream-Aware Tool Calling (SATC) enables the model to invoke external tools using the current visual and conversational context and incorporate returned information into the ongoing interaction. Interactive Visual Grounding (IVG) supports visual grounding in both user inputs and model outputs. The model can identify objects marked by users on video frames and produce frame-aligned bounding boxes to link its responses to specific objects and regions. To develop these capabilities, we construct a dedicated streaming interaction dataset and adopt a three-stage training scheme to integrate the new behaviors with AURA's existing streaming interaction abilities. We further incorporate Long-Horizon Interaction Memory (LHIM) to maintain contextual continuity over extended video streams. Evaluations on public streaming benchmarks show that AURA 2.0 achieves leading performance among open-source streaming models, while targeted evaluations further validate its new interaction capabilities. We open-source the model weights and inference code.

## How to share customer context across chat, SMS, and voice

DevFeed: [How to share customer context across chat, SMS, and voice](<https://devfeed.tech/articles/how-to-share-customer-context-across-chat-sms-and-voice-133765.md>)

Original publisher: [Read original article](<https://www.twilio.com/en-us/blog/insights/omnichannel-customer-context>)

Author: Jesse Sumrak

Published: 2026-11-13T00:00:00Z

Content type: tutorial

Language: en

Sources: [Twilio](<https://devfeed.tech/sources/twilio-blog.md>), [Twilio Blog](<https://devfeed.tech/sources/twilio-blog-2.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [Entity resolution](<https://devfeed.tech/topics/entity-resolution.md>), [AI for customer service automation](<https://devfeed.tech/topics/ai-for-customer-service-automation.md>), [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [chat](<https://devfeed.tech/tags/chat.md>), [contact-center](<https://devfeed.tech/tags/contact-center.md>), [context](<https://devfeed.tech/tags/context.md>), [conversation-memory](<https://devfeed.tech/tags/conversation-memory.md>), [conversation-orchestrator](<https://devfeed.tech/tags/conversation-orchestrator.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [conversations-api](<https://devfeed.tech/tags/conversations-api.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customers](<https://devfeed.tech/tags/customers.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [industry-insights](<https://devfeed.tech/tags/industry-insights.md>), [repeat](<https://devfeed.tech/tags/repeat.md>), [share](<https://devfeed.tech/tags/share.md>), [sms](<https://devfeed.tech/tags/sms.md>), [switch](<https://devfeed.tech/tags/switch.md>), [voice](<https://devfeed.tech/tags/voice.md>)

### AI overview

Twilio's approach depends on resolving every channel identifier to the same customer profile, then grouping conversations by that profile. Conversation Memory maps identifiers such as phone numbers, chat IDs, WhatsApp handles, and email to profiles; customers' alternate numbers must also be linked. Conversation Orchestrator uses profile grouping (GROUP_BY_PROFILE, which must be set explicitly) when resolution succeeds. Without a match, conversations fall back to grouping by address and do not connect to cross-channel history. Agent Connect can retrieve stored observations, summaries, and history before messages reach the agent, but its memory mode defaults to never and must be set to always or once. For chat-to-call handoffs that happen before a chat closes, save key facts as observations as they arise or use extraction on INACTIVE; otherwise the profile may not yet contain them. Human handoffs can pass profile and conversation IDs to Flex, where Unified Profiles shows customer history and Agent Copilot adds generated highlights; both are described as public beta.

### Source excerpt

Learn how to share customer context across chat, SMS, and voice, so customers don't repeat themselves when they switch channels.

[Next page](<https://devfeed.tech/topics/conversational-ai.md?cursor=WyIyMDI2LTExLTEzVDAwOjAwOjAwKzAwOjAwIiwgImY0MGMyYmQ5LTE5MTItNGY4Mi1hYTE2LWRkZTZhNWQ4NjcwMiJd>)