# human-in-the-loop UX

A user-experience pattern in which AI proposes or stages actions for human review and approval before execution.

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

## MATE: Mobile Agent Trustworthy Execution for Safe and Efficient Mobile Control

DevFeed: [MATE: Mobile Agent Trustworthy Execution for Safe and Efficient Mobile Control](<https://devfeed.tech/articles/mate-mobile-agent-trustworthy-execution-for-safe-and-efficient-mobile-control-145527.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.25526>)

Author: Minsuk Jang, Hee-Seon Kim, Hui-Cheol Yu, Changick Kim

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

Content type: article

Language: en

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

Topics: [computer-use](<https://devfeed.tech/topics/computer-use.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [block](<https://devfeed.tech/tags/block.md>), [blocking](<https://devfeed.tech/tags/blocking.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [data](<https://devfeed.tech/tags/data.md>), [decision](<https://devfeed.tech/tags/decision.md>), [device](<https://devfeed.tech/tags/device.md>), [effects](<https://devfeed.tech/tags/effects.md>), [execution](<https://devfeed.tech/tags/execution.md>), [fields](<https://devfeed.tech/tags/fields.md>), [gpt](<https://devfeed.tech/tags/gpt.md>), [guardrail](<https://devfeed.tech/tags/guardrail.md>), [interface](<https://devfeed.tech/tags/interface.md>), [layer](<https://devfeed.tech/tags/layer.md>), [llm](<https://devfeed.tech/tags/llm.md>), [mate](<https://devfeed.tech/tags/mate.md>), [mobile](<https://devfeed.tech/tags/mobile.md>), [model](<https://devfeed.tech/tags/model.md>), [safe](<https://devfeed.tech/tags/safe.md>)

### AI overview

MATE adds an execution layer between a mobile agent's proposed actions and their effects on the device, without modifying the underlying model. It records each UI action with its instruction, target app, screen context, primitive, and action fields. Runtime Guardrail uses that record to allow, block, or request user approval; Predefined Task Routing handles recurring requests through verified routes and checks the screen afterward. A compact floating overlay hosts execution. In reported evaluations, MATE increased high-risk blocking from 29.7% to 97.0% on MobileSafetyBench with GPT-4o; on AndroidWorld, it cut LLM calls by up to 6x and achieved 95.7% success on recurring-use tasks.

### Source excerpt

arXiv:2611.25526v1 Announce Type: cross Abstract: Mobile agents on personal smartphones can affect private data, payments, settings, and the foreground screen. We present MATE (Mobile Agent Trustworthy Execution), an execution-layer system that controls the boundary between agent action proposals and device effects without modifying the underlying model. MATE represents each UI primitive as a structured action record that combines the instruction, target app, screen context, primitive, and action fields. Runtime Guardrail (RG) maps this record to an allow, block, or user-approval decision; Predefined Task Routing (PTR) serves recurring requests through verified routes with post-execution screen checks; and a floating interface keeps execution in a compact overlay. On MobileSafetyBench, MATE raises high-risk blocking from 29.7% to 97.0% under GPT-4o. On AndroidWorld, it reduces LLM calls by up to 6x and raises recurring-use task success to 95.7%.

## Best Platforms for AI Agent Workflow Approval and Human Review in 2026

DevFeed: [Best Platforms for AI Agent Workflow Approval and Human Review in 2026](<https://devfeed.tech/articles/best-platforms-for-ai-agent-workflow-approval-and-human-review-in-2026-141049.md>)

Original publisher: [Read original article](<https://www.lyzr.ai/blog/best-platforms-ai-agent-workflow-approval-human-review/>)

Author: Lyzr Team

Published: 2026-11-18T08:10:04Z

Content type: comparison

Language: en

Sources: [Lyzr.ai](<https://devfeed.tech/sources/lyzr-ai.md>)

Topics: [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [n8n](<https://devfeed.tech/topics/n8n.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [approval-workflows](<https://devfeed.tech/tags/approval-workflows.md>), [human-review](<https://devfeed.tech/tags/human-review.md>), [platforms](<https://devfeed.tech/tags/platforms.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

AI approval platforms differ by what they pause or govern: Lyzr Agent Studio controls agent version promotion and production releases, while its documented flows do not provide a universal inbox for runtime actions. n8n and Zapier connect approval checkpoints to business apps; n8n can pause selected AI tool calls, while Zapier is aimed at simpler business-led automations. LangGraph provides stateful interrupts and resume controls but leaves reviewer interfaces and notifications to developers. Temporal adds durable, recoverable execution for long-running approvals, with approval interfaces and business logic also left to teams. Microsoft Copilot Studio supports multistage approvals through Teams and Outlook and fits organizations using Microsoft tools. The text recommends evaluating checkpoints, reviewer permissions, execution pause behavior, auditability, and deployment and governance needs; the best fit depends on whether approval applies to an action, an output, or production release.

### Source excerpt

An AI agent can process an invoice, draft a customer response, update a CRM record, or initiate a payment. But should it complete every action...

## Spara's Scout AI assistant acts on sales opportunities before reps know to look

DevFeed: [Spara's Scout AI assistant acts on sales opportunities before reps know to look](<https://devfeed.tech/articles/spara-s-scout-ai-assistant-acts-on-sales-opportunities-before-reps-know-to-look-137729.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/11/17/sparas-scout-ai-assistant-acts-on-sales-opportunities-before-reps-know-to-look/>)

Author: Duncan Riley

Published: 2026-11-17T14:00:50Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [apps](<https://devfeed.tech/tags/apps.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [chris-brown](<https://devfeed.tech/tags/chris-brown.md>), [context](<https://devfeed.tech/tags/context.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [crm](<https://devfeed.tech/tags/crm.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-relationship-management](<https://devfeed.tech/tags/customer-relationship-management.md>), [data](<https://devfeed.tech/tags/data.md>), [david-walker](<https://devfeed.tech/tags/david-walker.md>), [go-to-market](<https://devfeed.tech/tags/go-to-market.md>), [inspired-capital](<https://devfeed.tech/tags/inspired-capital.md>), [justworks](<https://devfeed.tech/tags/justworks.md>), [know](<https://devfeed.tech/tags/know.md>), [look](<https://devfeed.tech/tags/look.md>), [mindbody](<https://devfeed.tech/tags/mindbody.md>), [news](<https://devfeed.tech/tags/news.md>), [outreach](<https://devfeed.tech/tags/outreach.md>), [personalized-outreach](<https://devfeed.tech/tags/personalized-outreach.md>), [radical-ventures](<https://devfeed.tech/tags/radical-ventures.md>), [rho-technologies](<https://devfeed.tech/tags/rho-technologies.md>), [sales](<https://devfeed.tech/tags/sales.md>), [sales-opportunities](<https://devfeed.tech/tags/sales-opportunities.md>), [salesforce](<https://devfeed.tech/tags/salesforce.md>), [salespeople](<https://devfeed.tech/tags/salespeople.md>), [scout](<https://devfeed.tech/tags/scout.md>), [seed-funding](<https://devfeed.tech/tags/seed-funding.md>), [service](<https://devfeed.tech/tags/service.md>), [slack](<https://devfeed.tech/tags/slack.md>), [spara](<https://devfeed.tech/tags/spara.md>), [startup](<https://devfeed.tech/tags/startup.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>), [zander-pease](<https://devfeed.tech/tags/zander-pease.md>)

### AI overview

Spara's Scout monitors customer conversations, emails and CRM activity alongside signals such as job changes, company news and buying interest to identify sales opportunities and choose whom to contact. It sends findings through Slack and text, learns account context and seller preferences over time, drafts outreach in the seller's voice, and can join calls to contribute in real time; sellers retain control over what messages are sent. Spara reports that early users sent eight times as many personalized emails in their first two weeks, earned a 10.5% reply rate versus 3.5% for traditional sequences, and booked 29% more meetings. These are company-reported early results.

### Source excerpt

Artificial intelligence sales agent startup Spara Inc. today introduced Scout, an AI assistant designed to act on sales opportunities before salespeople know to look for them. Working around the clock, the new service analyzes context from customer conversations, emails and customer relationship management data to identify opportunities, prepare personalized outreach and complete work before reps [...] The post Spara's Scout AI assistant acts on sales opportunities before reps know to look appeared first on SiliconANGLE.

## Introducing eve 1.0

DevFeed: [Introducing eve 1.0](<https://devfeed.tech/articles/introducing-eve-1-0-138548.md>)

Original publisher: [Read original article](<https://vercel.com/blog/eve-is-now-generally-available>)

Author: Chad Hietala

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

Content type: release

Language: en

Sources: [Vercel News](<https://devfeed.tech/sources/vercel-news.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [Multi-step AI processes](<https://devfeed.tech/topics/multi-step-ai-processes.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>), [observability](<https://devfeed.tech/topics/observability.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [AI Sandbox](<https://devfeed.tech/topics/ai-sandbox.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-framework](<https://devfeed.tech/tags/agent-framework.md>), [agents](<https://devfeed.tech/tags/agents.md>), [build](<https://devfeed.tech/tags/build.md>), [delegates](<https://devfeed.tech/tags/delegates.md>), [durable-execution](<https://devfeed.tech/tags/durable-execution.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [eve](<https://devfeed.tech/tags/eve.md>), [faster](<https://devfeed.tech/tags/faster.md>), [framework](<https://devfeed.tech/tags/framework.md>), [harness](<https://devfeed.tech/tags/harness.md>), [hosting](<https://devfeed.tech/tags/hosting.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [integrations](<https://devfeed.tech/tags/integrations.md>), [multiplayer](<https://devfeed.tech/tags/multiplayer.md>), [observability](<https://devfeed.tech/tags/observability.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [production](<https://devfeed.tech/tags/production.md>), [sdk](<https://devfeed.tech/tags/sdk.md>), [security-controls](<https://devfeed.tech/tags/security-controls.md>)

### AI overview

eve 1.0 is a generally available open-source framework for building production agents, with a configurable harness for models, tools, instructions, approvals, and subagents. Its Workflow SDK saves progress for recovery after crashes and supports resumable tasks; managed Vercel hosting uses Workflows for durable execution and Sandbox for isolated code. The release adds shared multiplayer sessions across channels, tenant-specific credentials and policies, configurable cross-session memory, and OpenTelemetry tracing with controls for collection and redaction. The post reports a hosted 100-turn test falling from 3.036 seconds median per turn to 414 ms, excluding model generation, and estimates a Vercel Workflows cost reduction from 5.4c to 1.3c for a 50-turn session under stated assumptions. It also describes UI integrations, shared extensions, and a coding extension. Existing users are directed to follow the upgrade guide before deploying.

### Source excerpt

Today we're announcing eve 1.0, the open-source framework for building production-ready agents. Since the public preview in June, we've worked with early adopters to prepare eve for 1.0, shipping 230+ releases and merging thousands of pull requests. With this release, eve is now generally available. We've made eve faster, expanded integrations, and improved observability. Security controls let you decide which actions need approval and what data gets collected. We've also made it easier for coding agents to build with, measuring that experience with a rigorous evaluation suite. With eve handling your harness and infrastructure, you and your agent can focus on building instead. Build production agents on a harness you own The harness in eve is the loop that calls models, runs tools, and delegates work to subagents. You choose the model, add or replace tools, edit instructions, and decide which actions need approval. The open-source Workflow SDK saves progress as the agent runs, so it can pick up from its last completed step after a crash. Multiplayer sessions and agent-to-agent calls are built in. For managed hosting, deploy your eve agent to Vercel. Vercel Workflows handles durable execution, and Vercel Sandbox runs agent-generated code in isolation. Build software factories that turn requests into tested pull requests, company brains that answer questions from your business data, or customer-facing agents in your product. Thousands of businesses already run eve in production for internal and customer-facing agents. Rillet built a software factory on eve that takes in customer requests, converts them into specs, tests the changes, and opens PRs for human review. Rillet tripled its shipping rate in three months. Clay built monty, a data agent that answers employees' questions in Slack. Just over two months in, 30-40% of the company used it every week. Lindy's agent framework roundup names eve "Best for durable backend agents". Collaborate in multiplayer agent session

## Article: Governed Autonomous Intelligence for Mission-Critical Utility

DevFeed: [Article: Governed Autonomous Intelligence for Mission-Critical Utility](<https://devfeed.tech/articles/article-governed-autonomous-intelligence-for-mission-critical-utility-135325.md>)

Original publisher: [Read original article](<https://www.infoq.com/articles/utility-governed-autonomous-intelligence-ai/>)

Author: Purnang Borah

Published: 2026-11-16T11:00:00Z

Content type: article

Language: en

Sources: [InfoQ](<https://devfeed.tech/sources/infoq.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>), [audit](<https://devfeed.tech/topics/audit.md>), [Amazon Bedrock Knowledge Bases](<https://devfeed.tech/topics/amazon-bedrock-knowledge-bases.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-ai-architecture](<https://devfeed.tech/tags/agentic-ai-architecture.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-architecture](<https://devfeed.tech/tags/ai-architecture.md>), [application-performance-management](<https://devfeed.tech/tags/application-performance-management.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [architecture-design](<https://devfeed.tech/tags/architecture-design.md>), [article](<https://devfeed.tech/tags/article.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [audit](<https://devfeed.tech/tags/audit.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [autonomy](<https://devfeed.tech/tags/autonomy.md>), [aws](<https://devfeed.tech/tags/aws.md>), [but](<https://devfeed.tech/tags/but.md>), [control](<https://devfeed.tech/tags/control.md>), [covers](<https://devfeed.tech/tags/covers.md>), [critical](<https://devfeed.tech/tags/critical.md>), [enterprise-architecture](<https://devfeed.tech/tags/enterprise-architecture.md>), [execution](<https://devfeed.tech/tags/execution.md>), [gate](<https://devfeed.tech/tags/gate.md>), [governance](<https://devfeed.tech/tags/governance.md>), [intelligence](<https://devfeed.tech/tags/intelligence.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [mission](<https://devfeed.tech/tags/mission.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [observability](<https://devfeed.tech/tags/observability.md>), [operating-systems](<https://devfeed.tech/tags/operating-systems.md>), [operational-intelligence](<https://devfeed.tech/tags/operational-intelligence.md>), [operations-management](<https://devfeed.tech/tags/operations-management.md>), [performance](<https://devfeed.tech/tags/performance.md>), [utility](<https://devfeed.tech/tags/utility.md>), [utility-governed-autonomous-intelligence-ai](<https://devfeed.tech/tags/utility-governed-autonomous-intelligence-ai.md>)

### AI overview

The architecture keeps AI reasoning separate from production authority: agents correlate signals, retrieve runbooks, assess business impact, and propose remediation, while a deterministic, versioned action catalog sets risk and approval requirements. A governed tool gateway is the sole execution path, enforcing least privilege, checking authorization and approvals, validating arguments, recording evidence, and supporting post-action health checks. Prediction carries evidence and confidence but grants no authority. The AWS reference uses Bedrock AgentCore, Bedrock Knowledge Bases, and MCP; Azure and Google Cloud equivalents are also described, with some Azure governance features noted as preview. The platform can support compliance evidence but does not itself establish compliance. Autonomy starts read-only and expands by action class based on operational and governance evidence, with revocation available.

### Source excerpt

The article describes an architecture for agentic AI in regulated utility operations. Agents reason and propose actions, but a deterministic policy gate and a governed tool gateway control execution, with human approval where required. It covers an AWS reference implementation, grounding in runbooks, governed prediction, outcome validation, audit evidence, and phased autonomy. By Purnang Borah

## daily 2.0: The Kitchen, Picnic, and the Human Money Loop

DevFeed: [daily 2.0: The Kitchen, Picnic, and the Human Money Loop](<https://devfeed.tech/articles/daily-2-0-the-kitchen-picnic-and-the-human-money-loop-136356.md>)

Original publisher: [Read original article](<https://philipptheserver.com/posts/daily-2-0-the-kitchen-picnic-and-the-human-money-loop/>)

Author: Philipp Lehmann (philipp.lehmann@gruppe.ai)

Published: 2026-11-16T09:00:00Z

Content type: article

Language: en

Sources: [Philipp Lehmann](<https://devfeed.tech/sources/philipp-lehmann.md>)

Topics: [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [human review](<https://devfeed.tech/topics/human-review.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [angular](<https://devfeed.tech/tags/angular.md>), [api](<https://devfeed.tech/tags/api.md>), [app](<https://devfeed.tech/tags/app.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [automation](<https://devfeed.tech/tags/automation.md>), [checkout](<https://devfeed.tech/tags/checkout.md>), [cooking](<https://devfeed.tech/tags/cooking.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [daily](<https://devfeed.tech/tags/daily.md>), [fastapi](<https://devfeed.tech/tags/fastapi.md>), [kitchen](<https://devfeed.tech/tags/kitchen.md>), [money](<https://devfeed.tech/tags/money.md>), [python](<https://devfeed.tech/tags/python.md>), [shopping](<https://devfeed.tech/tags/shopping.md>), [stock](<https://devfeed.tech/tags/stock.md>)

### AI overview

daily 2.0 connects meal planning, physical pantry stock, cooking, shopping and Picnic deliveries while keeping payment and receipt decisions with the owner. It tracks home inventory as dated lots, records waste, and counts cooking only after confirmation; prepared portions are then consumed separately from raw ingredients. Shopping subtracts usable stock and trustworthy incoming quantities, while unresolved orders remain visible without being treated as either absent or received. Units stay explicit, and conversions require sourced quantities or approved rules. For Picnic, the system supports product search and reviewed draft carts, but sending requires an explicit command and checkout stays in the Picnic app. Delivery summaries are evidence rather than proof of received quantities, so receipts are confirmed per line. The unofficial, changeable API means uncertain cart writes must be inspected before any explicit recovery; live German-account compatibility still needs validation.

### Source excerpt

daily 2.0 grows a kitchen: physical expiring stock, confirmed cooking, stock-aware shopping, and approval-first Picnic carts -- I still do the checkout.

## Your next AI interface might just be a button, says PLAUD's Human-Computer Interaction Designer

DevFeed: [Your next AI interface might just be a button, says PLAUD's Human-Computer Interaction Designer](<https://devfeed.tech/articles/your-next-ai-interface-might-just-be-a-button-says-plaud-s-human-computer-interaction-designer-134645.md>)

Original publisher: [Read original article](<https://www.yankodesign.com/2026/11/15/your-next-ai-interface-might-just-be-a-button-says-plauds-human-computer-interaction-designer/>)

Author: Sarang Sheth

Published: 2026-11-16T02:45:11Z

Content type: article

Language: en

Sources: [Yanko Design](<https://devfeed.tech/sources/yanko-design.md>)

Topics: [human computer interaction](<https://devfeed.tech/topics/human-computer-interaction.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>), [User Experience](<https://devfeed.tech/topics/user-experience.md>), [Interaction Design](<https://devfeed.tech/topics/interaction-design.md>), [human centric design](<https://devfeed.tech/topics/human-centric-design.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [adfree](<https://devfeed.tech/tags/adfree.md>), [ai](<https://devfeed.tech/tags/ai.md>), [app](<https://devfeed.tech/tags/app.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-podcast-product-design-adfree-design-mindset-plaud](<https://devfeed.tech/tags/artificial-intelligence-podcast-product-design-adfree-design-mindset-plaud.md>), [attention](<https://devfeed.tech/tags/attention.md>), [button](<https://devfeed.tech/tags/button.md>), [capture](<https://devfeed.tech/tags/capture.md>), [computer](<https://devfeed.tech/tags/computer.md>), [design-mindset](<https://devfeed.tech/tags/design-mindset.md>), [designer](<https://devfeed.tech/tags/designer.md>), [human-computer-interaction](<https://devfeed.tech/tags/human-computer-interaction.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [interface](<https://devfeed.tech/tags/interface.md>), [notification](<https://devfeed.tech/tags/notification.md>), [phone](<https://devfeed.tech/tags/phone.md>), [plaud](<https://devfeed.tech/tags/plaud.md>), [podcast](<https://devfeed.tech/tags/podcast.md>), [product-design](<https://devfeed.tech/tags/product-design.md>)

### AI overview

The interview's central point is that AI interfaces can support attention through simple, legible physical controls, while more detailed review remains suited to a phone or computer. Jeff Easter describes a push-to-talk button for voice transcription and a brief press on PLAUD's recorder to mark important moments, giving AI human guidance instead of asking it to infer every priority. He emphasizes clear feedback about device state and visible signals that help people nearby understand when recording starts or stops; a vibration alone may inform the wearer without informing others. Captured transcripts and summaries should fit existing workflows, with confirmation before consequential actions such as sending messages. Screen-free design does not eliminate social or privacy concerns, and the interview's face-down/face-up recording totem is a hypothetical concept, not an announced PLAUD product.

### Source excerpt

Your next AI interface might just be a button, says PLAUD's Human-Computer Interaction Designer Pulling out a phone to capture an important thought can be the quickest way to lose it. A notification appears, another app demands attention, and...

## Introducing ChatGPT Wallet

DevFeed: [Introducing ChatGPT Wallet](<https://devfeed.tech/articles/introducing-chatgpt-wallet-133625.md>)

Original publisher: [Read original article](<https://openai.com/index/introducing-chatgpt-wallet>)

Published: 2026-11-13T15:00:00Z

Content type: release

Language: en

Sources: [OpenAI News](<https://devfeed.tech/sources/openai-news.md>)

Topics: [digital wallet](<https://devfeed.tech/topics/digital-wallet.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [agentic browsing](<https://devfeed.tech/topics/agentic-browsing.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [adapting](<https://devfeed.tech/tags/adapting.md>), [adyen](<https://devfeed.tech/tags/adyen.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-commerce](<https://devfeed.tech/tags/agentic-commerce.md>), [agents](<https://devfeed.tech/tags/agents.md>), [american-express](<https://devfeed.tech/tags/american-express.md>), [auth](<https://devfeed.tech/tags/auth.md>), [card](<https://devfeed.tech/tags/card.md>), [chatgpt](<https://devfeed.tech/tags/chatgpt.md>), [chatgpt-wallet](<https://devfeed.tech/tags/chatgpt-wallet.md>), [chatgpt-work](<https://devfeed.tech/tags/chatgpt-work.md>), [codex](<https://devfeed.tech/tags/codex.md>), [complete](<https://devfeed.tech/tags/complete.md>), [dots](<https://devfeed.tech/tags/dots.md>), [existing](<https://devfeed.tech/tags/existing.md>), [product](<https://devfeed.tech/tags/product.md>), [user](<https://devfeed.tech/tags/user.md>), [wallet](<https://devfeed.tech/tags/wallet.md>), [work](<https://devfeed.tech/tags/work.md>)

### AI overview

ChatGPT Wallet lets an agent pay with a user's selected card while withholding the card's actual details from the agent and business. Before each purchase, the user reviews the items, total, and card, then approves with a security check such as Face ID or fingerprint; the payment network supplies a token, and the user receives a record of the authorization. Eligible card rewards and dispute protections remain in place, while businesses retain control of pricing, fulfillment, returns, and customer service. The initial rollout is through dots for ChatGPT Pro subscribers aged 18 or older in the US with eligible US-issued cards. Businesses can update verified product details and feeds, allow ChatGPT agents through site security using signed Web Bot Auth requests, and work with payment and risk providers to reduce declines. The article says token payments fit existing checkout flows, though risk systems may still flag legitimate agent purchases. Future delegated purchases with limits and pre-approval are described as an expectation, not a current feature.

### Source excerpt

Introducing ChatGPT Wallet, a secure way for agents in ChatGPT Work, Codex, and dots to complete purchases with user approval and existing card protections.

## AI Automation Safety: When to Keep a Human in the Loop

DevFeed: [AI Automation Safety: When to Keep a Human in the Loop](<https://devfeed.tech/articles/ai-automation-safety-when-to-keep-a-human-in-the-loop-130182.md>)

Original publisher: [Read original article](<https://hackernoon.com/ai-automation-safety-when-to-keep-a-human-in-the-loop?source=rss>)

Author: Muneeb Ashraf

Published: 2026-11-12T07:33:50Z

Content type: tutorial

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [workflow automation](<https://devfeed.tech/topics/workflow-automation.md>), [AI for customer service automation](<https://devfeed.tech/topics/ai-for-customer-service-automation.md>), [Automation](<https://devfeed.tech/topics/automation.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-automation](<https://devfeed.tech/tags/ai-automation.md>), [ai-automation-agency](<https://devfeed.tech/tags/ai-automation-agency.md>), [ai-customer-support](<https://devfeed.tech/tags/ai-customer-support.md>), [ai-safety](<https://devfeed.tech/tags/ai-safety.md>), [ai-workflow](<https://devfeed.tech/tags/ai-workflow.md>), [airtable](<https://devfeed.tech/tags/airtable.md>), [audit-trail](<https://devfeed.tech/tags/audit-trail.md>), [automation](<https://devfeed.tech/tags/automation.md>), [customer-support](<https://devfeed.tech/tags/customer-support.md>), [dashboard](<https://devfeed.tech/tags/dashboard.md>), [dutch](<https://devfeed.tech/tags/dutch.md>), [human-in-the-loop](<https://devfeed.tech/tags/human-in-the-loop.md>), [invoice-automation](<https://devfeed.tech/tags/invoice-automation.md>), [workflow-automation](<https://devfeed.tech/tags/workflow-automation.md>), [zapier](<https://devfeed.tech/tags/zapier.md>)

### AI overview

The article recommends keeping human approval on actions that are hard to reverse, leave the organization's control, or carry costs that outweigh a short delay. Its examples include automatically answering tested routine support questions while routing unclear or money-related cases to reviewers, and holding invoices when bank details differ from supplier records until someone verifies the change through a contact already on file. It also describes silent workflow failure despite an "active" status, recommending checks for expected output, and treats retired-model replacements as new releases requiring about three weeks of testing across languages, outputs, and human-reviewed live runs. In Zapier, Human in the Loop pauses a Zap for a reviewer's decision; later steps can use the decision and reviewer note, with approvals and declines logged. The article says the feature requires a paid plan, reviewers need Zapier accounts and shared access, Pro allows requests only to yourself, and the step cannot be inside or after a Looping by Zapier step.

### Source excerpt

AI can send routine replies, but it never pays or deletes. How we decide which automation steps need a human approval, and how to build one in Zapier.

## How Red Hat OpenShift 5 simplifies incident diagnosis with supervised AI

DevFeed: [How Red Hat OpenShift 5 simplifies incident diagnosis with supervised AI](<https://devfeed.tech/articles/how-red-hat-openshift-5-simplifies-incident-diagnosis-with-supervised-ai-129635.md>)

Original publisher: [Read original article](<https://www.redhat.com/en/blog/how-red-hat-openshift-5-simplifies-incident-diagnosis-supervised-ai>)

Author: Vanessa Martini

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

Content type: article

Language: en

Sources: [Red Hat Blog](<https://devfeed.tech/sources/red-hat-blog.md>)

Topics: [AIOps for Kubernetes](<https://devfeed.tech/topics/aiops-for-kubernetes.md>), [observability](<https://devfeed.tech/topics/observability.md>), [telemetry](<https://devfeed.tech/topics/telemetry.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [Prometheus](<https://devfeed.tech/topics/prometheus.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [alert](<https://devfeed.tech/tags/alert.md>), [alert-fatigue](<https://devfeed.tech/tags/alert-fatigue.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [data](<https://devfeed.tech/tags/data.md>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [engineer](<https://devfeed.tech/tags/engineer.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [incident](<https://devfeed.tech/tags/incident.md>), [language](<https://devfeed.tech/tags/language.md>), [logs](<https://devfeed.tech/tags/logs.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [openshift](<https://devfeed.tech/tags/openshift.md>), [openshift-5](<https://devfeed.tech/tags/openshift-5.md>), [platform](<https://devfeed.tech/tags/platform.md>), [prometheus](<https://devfeed.tech/tags/prometheus.md>), [red-hat-openshift](<https://devfeed.tech/tags/red-hat-openshift.md>), [security-governance](<https://devfeed.tech/tags/security-governance.md>), [telemetry](<https://devfeed.tech/tags/telemetry.md>), [traces](<https://devfeed.tech/tags/traces.md>)

### AI overview

OpenShift Lightspeed turns alert response into structured, human-reviewed investigation: it can correlate Prometheus metrics, pod logs, and installed OpenTelemetry traces into evidence, signals, and draft findings. On OpenShift 4.22+ with Lightspeed operator 1.1.5+, teams can also request Perses metric panels in natural language; with cluster observability operator 1.6, panels are marked AI-assisted and expose PromQL, time range, and source metrics for review. OpenShift 5 adds an opt-in technology preview that watches existing Alertmanager receivers, runs standard oc/kubectl diagnostics, and stages remediation for approval. It requires Lightspeed operator 2 and a configured tool-calling LLM; no default model ships. Changes require administrator approval, run in ephemeral RBAC-sandboxed environments, and are audited, with failed remediation halted and escalated. Model endpoint choice determines where data is processed; locally hosted models can support air-gapped use.

### Source excerpt

Prometheus metrics, traces, and logs are rarely missing in modern enterprise environments. Most platform teams already have plenty of telemetry at their disposal. The real bottleneck is the sheer volume of data a human must parse before taking meaningful action during an incident.On a busy cluster, this creates a familiar and exhausting pattern: an alert fires, an engineer opens 3 different consoles, brings up a Prometheus Query Language (PromQL) scratchpad, executes a log query, and spends 20 minutes debating whether the signal represents a genuine emergency or background noise. Alert fatigue

## AI security startup Bay launches with $50M to vet agent actions before they run

DevFeed: [AI security startup Bay launches with $50M to vet agent actions before they run](<https://devfeed.tech/articles/ai-security-startup-bay-launches-with-50m-to-vet-agent-actions-before-they-run-126348.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/11/10/ai-security-startup-bay-launches-with-50m-to-vet-agent-actions-before-they-run/>)

Author: Duncan Riley

Published: 2026-11-10T13:00:13Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [ai security](<https://devfeed.tech/topics/ai-security.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [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-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-security](<https://devfeed.tech/tags/ai-security.md>), [aner-morag](<https://devfeed.tech/tags/aner-morag.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-security](<https://devfeed.tech/tags/artificial-intelligence-security.md>), [assaf-rappaport](<https://devfeed.tech/tags/assaf-rappaport.md>), [battery-ventures](<https://devfeed.tech/tags/battery-ventures.md>), [bay](<https://devfeed.tech/tags/bay.md>), [bay-ai](<https://devfeed.tech/tags/bay-ai.md>), [bay-enforcer](<https://devfeed.tech/tags/bay-enforcer.md>), [boris-danilovich](<https://devfeed.tech/tags/boris-danilovich.md>), [clickup](<https://devfeed.tech/tags/clickup.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [company](<https://devfeed.tech/tags/company.md>), [crowdstrike](<https://devfeed.tech/tags/crowdstrike.md>), [crowdstrike-holdings](<https://devfeed.tech/tags/crowdstrike-holdings.md>), [cyber-hunters](<https://devfeed.tech/tags/cyber-hunters.md>), [employee](<https://devfeed.tech/tags/employee.md>), [endpoint-detection-and-response](<https://devfeed.tech/tags/endpoint-detection-and-response.md>), [endpoint-security](<https://devfeed.tech/tags/endpoint-security.md>), [enterprises](<https://devfeed.tech/tags/enterprises.md>), [fortune-100](<https://devfeed.tech/tags/fortune-100.md>), [funding](<https://devfeed.tech/tags/funding.md>), [israel-defense-forces](<https://devfeed.tech/tags/israel-defense-forces.md>), [launches](<https://devfeed.tech/tags/launches.md>), [lighthouse](<https://devfeed.tech/tags/lighthouse.md>), [micron-technology](<https://devfeed.tech/tags/micron-technology.md>), [micron-ventures](<https://devfeed.tech/tags/micron-ventures.md>), [mobile-device-management](<https://devfeed.tech/tags/mobile-device-management.md>), [news](<https://devfeed.tech/tags/news.md>), [noname-security](<https://devfeed.tech/tags/noname-security.md>), [peloton-interactive](<https://devfeed.tech/tags/peloton-interactive.md>), [preempt-security](<https://devfeed.tech/tags/preempt-security.md>), [run](<https://devfeed.tech/tags/run.md>), [security](<https://devfeed.tech/tags/security.md>), [security-startup](<https://devfeed.tech/tags/security-startup.md>), [seed-round](<https://devfeed.tech/tags/seed-round.md>), [series-a](<https://devfeed.tech/tags/series-a.md>), [startup](<https://devfeed.tech/tags/startup.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>)

### AI overview

Bay says its platform checks AI agents' proposed actions before execution across laptops, cloud workloads, and SaaS, using one policy set. Its Wave scanner inventories agents, extensions, and local models, then stops running; Bay Enforcer receives each proposed action through a native hook and evaluates it on the machine using the session context. Risky actions are blocked, while some checks prompt users for approval. Lighthouse drafts team policies from the prior 30 days of activity, which teams can test against that month before use. Bay reports verdicts in milliseconds and false positives below 1%. The company says installation through existing MDM or EDR tools takes under five minutes. It reports dozens of enterprise customers, including Micron, Wix, ClickUp, and Peloton; these product and performance claims are attributed to Bay.

### Source excerpt

Israeli artificial intelligence security startup Bay AI Inc. launched today with $50 million in funding to let enterprises vet each action an AI agent tries to take before it runs. Bay provides software that tracks the agents working across a company, whether they sit on an employee's laptop, inside a cloud workload or in a [...] The post AI security startup Bay launches with $50M to vet agent actions before they run appeared first on SiliconANGLE.

## Claude Code's "don't ask again" button is leaking passwords and API keys to public GitHub repos

DevFeed: [Claude Code's "don't ask again" button is leaking passwords and API keys to public GitHub repos](<https://devfeed.tech/articles/claude-code-s-don-t-ask-again-button-is-leaking-passwords-and-api-keys-to-public-github-repos-125810.md>)

Original publisher: [Read original article](<https://www.techspot.com/news/114500-claude-code-dont-ask-again-button-leaking-passwords.html>)

Author: Skye Jacobs

Published: 2026-11-10T12:03:00Z

Content type: news

Language: en

Sources: [TechSpot](<https://devfeed.tech/sources/techspot.md>)

Topics: [Secret Scanning](<https://devfeed.tech/topics/secret-scanning.md>), [hardcoded credentials](<https://devfeed.tech/topics/hardcoded-credentials.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [api-keys](<https://devfeed.tech/tags/api-keys.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-ai](<https://devfeed.tech/tags/claude-ai.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [command](<https://devfeed.tech/tags/command.md>), [commands](<https://devfeed.tech/tags/commands.md>), [file](<https://devfeed.tech/tags/file.md>), [full](<https://devfeed.tech/tags/full.md>), [general](<https://devfeed.tech/tags/general.md>), [github](<https://devfeed.tech/tags/github.md>), [json](<https://devfeed.tech/tags/json.md>), [keys](<https://devfeed.tech/tags/keys.md>), [leak](<https://devfeed.tech/tags/leak.md>), [local](<https://devfeed.tech/tags/local.md>), [machine](<https://devfeed.tech/tags/machine.md>), [options](<https://devfeed.tech/tags/options.md>), [passwords](<https://devfeed.tech/tags/passwords.md>), [pattern](<https://devfeed.tech/tags/pattern.md>), [public](<https://devfeed.tech/tags/public.md>), [security](<https://devfeed.tech/tags/security.md>), [settings](<https://devfeed.tech/tags/settings.md>), [shell](<https://devfeed.tech/tags/shell.md>), [stored](<https://devfeed.tech/tags/stored.md>)

### AI overview

Choosing "Yes, and don't ask again" in Claude Code can save the exact shell command in `.claude/settings.local.json`, including any credentials it contains. Apiiro found this file in over 100,000 public repositories; about 1,100 contained secrets from 63 organizations. A global Git ignore rule may not protect a repository cloned elsewhere or created around an existing file, and the article says Claude Code sometimes commits settings changes itself. Removing a leaked file does not remove copies in existing forks. Gitleaks and TruffleHog missed 215 of the exposed secrets. Claude Code team member Boris Cherny said the team was looking into safeguards, pointed to auto mode for routine permission prompts, and said built-in secret scanning was under consideration; a commenter noted auto mode could still save sensitive commands. The article describes earlier bug reports that closed without a fix and argues that storing raw commands leaves credentials dependent on ignore rules working across repository setups.

### Source excerpt

Claude Code asks for approval before it runs shell commands on a developer's machine. One of the options is "Yes, and don't ask again." Choosing it creates an allow rule in a file called .claude/settings.local.json. Sometimes the rule is a general pattern, but often it's the full command, stored word... Read Entire Article

## From Delegation to Action Approval: An Exploratory Study of Developers' Practices in Collaborating with AI Coding Agents

DevFeed: [From Delegation to Action Approval: An Exploratory Study of Developers' Practices in Collaborating with AI Coding Agents](<https://devfeed.tech/articles/from-delegation-to-action-approval-an-exploratory-study-of-developers-practices-in-collaborating-with-ai-coding-agents-127223.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.10058>)

Author: Yinan Wu, Dominik Wermke, Ze Shi Li, Kathryn Thomasset Stolee, Bowen Xu

Published: 2026-11-10T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [developer-productivity](<https://devfeed.tech/topics/developer-productivity.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [autonomy](<https://devfeed.tech/tags/autonomy.md>), [code-review](<https://devfeed.tech/tags/code-review.md>), [coding](<https://devfeed.tech/tags/coding.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [delegation](<https://devfeed.tech/tags/delegation.md>), [developers](<https://devfeed.tech/tags/developers.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [security](<https://devfeed.tech/tags/security.md>)

### AI overview

A survey of 203 developers found that careful review of generated code did not ensure careful review of agent actions: 61% of those who carefully reviewed most or all generated code still often approved at least one type of action without close review. Developers aware of security or privacy incidents or concerning practices did so more often. Autonomy varied more between developers than between development activities. Respondents reported a lighter workload but said supervising agents remained mentally demanding. More than half of responses about granting agents greater autonomy called for bounded, observable, or reversible behavior, suggesting reliability alone is not enough.

### Source excerpt

arXiv:2611.10058v1 Announce Type: new Abstract: Developers now collaborate with AI coding agents that write code and take actions on their behalf. Beyond reviewing this code, developers decide what to delegate and which agent actions to approve. Some of these actions, such as running shell commands or reading files, can have effects that never appear in the resulting code. How developers make these decisions and how much effort this requires are not well understood. To better understand this collaboration, we surveyed 203 developers who use AI coding agents. We found that careful code review does not necessarily extend to action approval: 61% of developers who carefully reviewed most or all generated code still often approved at least one type of agent action without close review. Developers aware of security or privacy incidents or concerning practices did so more often than those who were not. We also found that autonomy varied more between developers than between development activities. Although developers reported a lighter workload, supervising agents remained mentally demanding. When asked what would make them grant more autonomy, developers wanted more than reliability: more than half of their responses called for agents that are bounded, observable, or reversible. Our findings suggest that as agents take on more of the coding, the challenge shifts toward supporting the developers who supervise them.

## Brain-Aligned Latent Actions: Imposing Premotor Cortex Geometry on a Robot's Action Vocabulary

DevFeed: [Brain-Aligned Latent Actions: Imposing Premotor Cortex Geometry on a Robot's Action Vocabulary](<https://devfeed.tech/articles/brain-aligned-latent-actions-imposing-premotor-cortex-geometry-on-a-robot-s-action-vocabulary-125426.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.11674>)

Author: Amit Jadhav

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

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

Topics: [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [VAE](<https://devfeed.tech/topics/vae.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [brain](<https://devfeed.tech/tags/brain.md>), [cortex](<https://devfeed.tech/tags/cortex.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [manipulation](<https://devfeed.tech/tags/manipulation.md>), [monkey](<https://devfeed.tech/tags/monkey.md>), [motion](<https://devfeed.tech/tags/motion.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [robot](<https://devfeed.tech/tags/robot.md>), [tokenizers](<https://devfeed.tech/tags/tokenizers.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

Latent action codes are highly unstable across training runs and even precision formats: independent Moto tokenizer runs produce an adjusted Rand index of 0.001, while bf16 evaluation reassigns 74% of tokens. The authors use this flexibility to train a tokenizer with a straight-through similarity loss that aligns its code geometry to macaque ventral premotor cortex (PMv), alongside the reconstruction objective. On five held-out shape categories, the PMv-trained tokenizer scores 0.52 representational similarity with PMv, versus 0.20-0.26 for released and no-prior tokenizers. However, a hand-picked initial-stroke-direction descriptor scores 0.65, and a label-shuffled PMv control scores 0.41; kinematic, language, and random targets do not transfer. The abstract says a downstream CALVIN manipulation test is specified before being run, so it provides no outcome for that test.

### Source excerpt

arXiv:2611.11674v1 Announce Type: cross Abstract: Robot policies increasingly act through a small vocabulary of discrete latent actions learned from unlabeled video, yet training leaves open which motion receives which code. We find this choice highly unstable: two independently trained runs of Moto's latent motion tokenizer assign the same robot clips to nearly unrelated codes (adjusted Rand index 0.001), and evaluating one tokenizer in bf16 instead of fp32 reassigns 74% of its tokens. We use this freedom to impose the representational geometry of macaque ventral premotor cortex (PMv), recorded while a monkey drew shapes, on the tokenizer's code assignment, with a straight-through similarity loss trained alongside its reconstruction objective on robot video. In one training run per target, on five shape categories the loss never saw, the PMv-trained tokenizer reaches a representational similarity of 0.52 with PMv, against 0.20-0.26 for released and no-prior tokenizers and 0.42 for the 95th percentile of random kinematic encoders. Kinematic, language, and random targets do not transfer (-0.17 to -0.03). Two results limit the claim: a hand-picked kinematic descriptor, the stroke's initial direction, matches PMv better (0.65), and a label-shuffled PMv control scores 0.41. We specify the downstream manipulation test, on a CALVIN pipeline that reproduces Moto's published results, before running it.

## Evidence-Linked Radiology Reporting: A Human-Supervised Reference Architecture for Structured Imaging Intelligence

DevFeed: [Evidence-Linked Radiology Reporting: A Human-Supervised Reference Architecture for Structured Imaging Intelligence](<https://devfeed.tech/articles/evidence-linked-radiology-reporting-a-human-supervised-reference-architecture-for-structured-imaging-intelligence-124427.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.25120>)

Author: Houman Kazemzadeh, Kamyar Naderi

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

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

Topics: [Medical imaging](<https://devfeed.tech/topics/medical-imaging.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-assisted](<https://devfeed.tech/tags/ai-assisted.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [comparison](<https://devfeed.tech/tags/comparison.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>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [governance](<https://devfeed.tech/tags/governance.md>), [imaging](<https://devfeed.tech/tags/imaging.md>), [interoperability](<https://devfeed.tech/tags/interoperability.md>), [measurement](<https://devfeed.tech/tags/measurement.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [regulatory](<https://devfeed.tech/tags/regulatory.md>), [reporting](<https://devfeed.tech/tags/reporting.md>), [speech](<https://devfeed.tech/tags/speech.md>), [structured](<https://devfeed.tech/tags/structured.md>), [templates](<https://devfeed.tech/tags/templates.md>), [validation](<https://devfeed.tech/tags/validation.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

The paper proposes a human-supervised structured reporting layer that links radiology findings to image evidence, measurements, prior comparisons, lesion identity, uncertainty, and terminology--information often fragmented across imaging and clinical systems. It combines exam-specific templates, speech-to-structure processing, measurement and segmentation capture, and controlled AI-assisted drafting, with interoperability based on DICOM, HL7 FHIR, and clinical terminology standards. The system is intended to support radiologist-reviewed reports, longitudinal comparison, data reuse, governance, and integration with PACS, RIS, EHR, analytics, and registries; it is explicitly not an autonomous report generator. The abstract also identifies modality-specific deployment, safety, validation, cybersecurity, privacy, quality management, and regulatory boundaries as considerations.

### Source excerpt

arXiv:2605.25120v2 Announce Type: replace-cross Abstract: Radiology reports remain the primary mechanism by which imaging findings are communicated to clinical teams. However, much of the structured information behind these reports, including measurements, image evidence, prior comparisons, lesion identity, uncertainty, and terminology, often remains trapped in free text or fragmented across picture archiving and communication systems, radiology information systems, reporting workstations, worksheets, advanced visualization tools, and electronic health records. This paper proposes a human-supervised, evidence-linked reference architecture for structured radiology reporting. The framework combines exam-specific templates, speech-to-structure processing, measurement and segmentation capture, controlled AI-assisted drafting, and standards-based interoperability using DICOM, DICOM Structured Reporting, DICOM Segmentation, HL7 FHIR, RadLex, SNOMED CT, LOINC, and UCUM. The system is positioned not as an autonomous report generator, but as a structured intelligence layer for enterprise imaging that supports reviewed reporting, longitudinal comparison, clinical data reuse, governance, and integration with PACS, RIS, EHR, analytics, and registry workflows. The paper also discusses modality-specific deployment considerations, clinical safety risks, validation requirements, cybersecurity, privacy, quality management, and regulatory boundaries for AI-assisted radiology reporting systems.

## AI Thinks. Humans Decide.

DevFeed: [AI Thinks. Humans Decide.](<https://devfeed.tech/articles/ai-thinks-humans-decide-122903.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/08/ai-thinks-humans-decide/>)

Author: Ritesh Patel

Published: 2026-11-08T20:00:00Z

Content type: opinion

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [ai-governance](<https://devfeed.tech/topics/ai-governance.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [accountability](<https://devfeed.tech/tags/accountability.md>), [actions](<https://devfeed.tech/tags/actions.md>), [admission](<https://devfeed.tech/tags/admission.md>), [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-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [airline](<https://devfeed.tech/tags/airline.md>), [amazon](<https://devfeed.tech/tags/amazon.md>), [api-key](<https://devfeed.tech/tags/api-key.md>), [automation](<https://devfeed.tech/tags/automation.md>), [aws](<https://devfeed.tech/tags/aws.md>), [blast-radius](<https://devfeed.tech/tags/blast-radius.md>), [chatbot](<https://devfeed.tech/tags/chatbot.md>), [ci](<https://devfeed.tech/tags/ci.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [decision](<https://devfeed.tech/tags/decision.md>), [human-in-the-loop](<https://devfeed.tech/tags/human-in-the-loop.md>), [incidents](<https://devfeed.tech/tags/incidents.md>), [policy-as-code](<https://devfeed.tech/tags/policy-as-code.md>)

### AI overview

AI can estimate outcomes and generate options, but it lacks both some decision-relevant context and accountability for whose interests or losses a choice accepts. The text argues that adding data cannot resolve the accountability gap: a person must own consequential tradeoffs. For agents that execute actions, prompts are not reliable controls; policy should be checked before execution, actions linked to both a human and agent identity, and policies observed in shadow mode before enforcement. It recommends letting agents handle reversible, low-cost work while named humans retain expensive, irreversible or ambiguous decisions. The article presents AIControls as an implementation of this approach, including use with Claude Code and a design partner's LiteLLM setup. It notes that two cited incidents have qualifications: Zillow had human inspectors, and Amazon disputes that AI caused its outage.

### Source excerpt

AI can think through a decision but can't own it. Why people set the limits for AI agents, four real incidents, and how to enforce them before an action runs.

## Aina's $399 AI Ring Turns Your Thumb Into an AI Approval Button

DevFeed: [Aina's $399 AI Ring Turns Your Thumb Into an AI Approval Button](<https://devfeed.tech/articles/aina-s-399-ai-ring-turns-your-thumb-into-an-ai-approval-button-121604.md>)

Original publisher: [Read original article](<https://www.techrepublic.com/article/news-aina-ai-ring-task-approvals/>)

Author: Aminu Abdullahi

Published: 2026-11-06T18:03:54Z

Content type: news

Language: en

Sources: [TechRepublic](<https://devfeed.tech/sources/techrepublic.md>)

Topics: [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [function calling](<https://devfeed.tech/topics/function-calling.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-ring](<https://devfeed.tech/tags/ai-ring.md>), [aina](<https://devfeed.tech/tags/aina.md>), [aina-ring](<https://devfeed.tech/tags/aina-ring.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [compatibility](<https://devfeed.tech/tags/compatibility.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [iphone](<https://devfeed.tech/tags/iphone.md>), [it](<https://devfeed.tech/tags/it.md>), [mac](<https://devfeed.tech/tags/mac.md>), [smart-rings](<https://devfeed.tech/tags/smart-rings.md>), [teams](<https://devfeed.tech/tags/teams.md>), [wearable-technology](<https://devfeed.tech/tags/wearable-technology.md>), [workplace-productivity](<https://devfeed.tech/tags/workplace-productivity.md>)

### AI overview

Aina's Ring is a companion for iPhone and Mac that uses a thumb flick to approve one suggested action at a time, alongside gesture controls for music and meetings and voice input for notes or prompts. Aina says it learns routines from location and calendar activity. It requires iOS 18 or later for iPhone use and macOS 15 or later for Mac features; Android and Windows are unsupported. Preorders start at $399, with shipments expected in early 2027, subject to destination availability. For work use, teams should verify which functions support their tools and require a connected AI agent, review each integration's data handling and permissions, and pilot routine tasks before enabling workflows involving messages, sensitive information, or spending. The excerpt notes that a faster approval is useful only if employees can understand the proposed action and verify its result.

### Source excerpt

Aina's $399 AI ring lets users approve tasks with a thumb flick. Check iPhone and Mac compatibility, shipping plans, and key considerations for IT teams. The post Aina's $399 AI Ring Turns Your Thumb Into an AI Approval Button appeared first on TechRepublic.

## Sysdig AI Defense puts runtime guardrails on AI agents

DevFeed: [Sysdig AI Defense puts runtime guardrails on AI agents](<https://devfeed.tech/articles/sysdig-ai-defense-puts-runtime-guardrails-on-ai-agents-119038.md>)

Original publisher: [Read original article](<https://www.helpnetsecurity.com/2026/11/05/sysdig-ai-defense/>)

Author: Industry News

Published: 2026-11-05T13:29:28Z

Content type: release

Language: en

Sources: [Help Net Security](<https://devfeed.tech/sources/help-net-security.md>)

Topics: [Falco](<https://devfeed.tech/topics/falco.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [shadow AI](<https://devfeed.tech/topics/shadow-ai.md>), [Endpoint Security & XDR](<https://devfeed.tech/topics/endpoint-security-xdr.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-defense](<https://devfeed.tech/tags/ai-defense.md>), [guardrails](<https://devfeed.tech/tags/guardrails.md>), [industry-news](<https://devfeed.tech/tags/industry-news.md>), [runtime](<https://devfeed.tech/tags/runtime.md>), [sysdig](<https://devfeed.tech/tags/sysdig.md>), [sysdig-ai-defense](<https://devfeed.tech/tags/sysdig-ai-defense.md>)

### AI overview

Sysdig AI Defense monitors AI agent activity at runtime by combining harness telemetry--which shows tools, MCP servers, skills, plugins, arguments, and targets--with kernel-level records of processes and actions actually executed. It evaluates actions against policy and the agent's current reach, allowing, blocking, or holding them for human review; it can also kill a process that gets past a block. Policies can start in alert mode, with templates mapped to OWASP's Top 10 for LLM Applications. The product records sessions and keeps immutable policy and activity evidence for investigation and audit, and attributes model calls to agents, teams, and people, including calls outside an AI gateway. Sysdig says it complements endpoint protection, model testing, and red teaming, and covers workstations, Linux and Kubernetes production environments, and managed agent platforms.

### Source excerpt

Sysdig has announced the general availability of Sysdig AI Defense, giving organizations tools to monitor and control AI agent activity across endpoints and cloud environments. The platform checks agent actions against company policies to allow them, hold them for user approval or block them before completion. It also records agent activity and policy decisions for investigations and audits. Enterprises are deploying AI faster than they can govern it. In a recent IBM survey, 77% of ... More -> The post Sysdig AI Defense puts runtime guardrails on AI agents appeared first on Help Net Security.

## Harness for Agents: Introducing a New Platform Experience

DevFeed: [Harness for Agents: Introducing a New Platform Experience](<https://devfeed.tech/articles/harness-for-agents-introducing-a-new-platform-experience-119403.md>)

Original publisher: [Read original article](<https://www.harness.io/blog/introducing-a-new-platform-experience>)

Author: Rohan Gupta

Published: 2026-11-05T00:00:00Z

Content type: release

Language: en

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

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [internal developer platform](<https://devfeed.tech/topics/internal-developer-platform.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [platform engineering automation](<https://devfeed.tech/topics/platform-engineering-automation.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-chat](<https://devfeed.tech/tags/ai-chat.md>), [audit](<https://devfeed.tech/tags/audit.md>), [audit-trail](<https://devfeed.tech/tags/audit-trail.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [build](<https://devfeed.tech/tags/build.md>), [builder](<https://devfeed.tech/tags/builder.md>), [containerized](<https://devfeed.tech/tags/containerized.md>), [context](<https://devfeed.tech/tags/context.md>), [control](<https://devfeed.tech/tags/control.md>), [copy](<https://devfeed.tech/tags/copy.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [deployments](<https://devfeed.tech/tags/deployments.md>), [experience](<https://devfeed.tech/tags/experience.md>), [harness](<https://devfeed.tech/tags/harness.md>), [platform](<https://devfeed.tech/tags/platform.md>)

### AI overview

Harness describes a platform experience designed for people and agents to author and run software delivery work through one pipeline model. Build, deploy, security, infrastructure, and agentic tasks use reusable steps in flatter YAML, with JEXL and CEL supported; converted OPA policies need updating and retesting before use. A unified Go Delegate orchestrates containerized steps, so deployment tools run with each step rather than on the Delegate host. Teams can review and pin versioned templates, while Worker Agents run within existing approvals, policies, permissions, and audit trails; user-facing agents propose actions for approval. Dashboards use the Harness Data Platform and Software Delivery Knowledge Graph, but coverage and metric parity are still being validated and Looker dashboards do not migrate automatically. Early access rolls out by capability; the old and new experiences run side by side, and pipeline conversion is optional and creates a copy.

### Source excerpt

Hundreds of millions of deployments taught us what delivery takes. The new platform experience applies it to a world where agents do the work. | Blog

## Smart Approvals in Amazon Quick: Human oversight where it matters

DevFeed: [Smart Approvals in Amazon Quick: Human oversight where it matters](<https://devfeed.tech/articles/smart-approvals-in-amazon-quick-human-oversight-where-it-matters-117740.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/machine-learning/smart-approvals-in-amazon-quick-human-oversight-where-it-matters/>)

Author: Steve Vinsik

Published: 2026-11-04T16:33:03Z

Content type: article

Language: en

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

Topics: [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [ai-governance](<https://devfeed.tech/topics/ai-governance.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [audit](<https://devfeed.tech/topics/audit.md>), [Amazon CloudWatch](<https://devfeed.tech/topics/amazon-cloudwatch.md>)

Tags: [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [amazon-cloudwatch-logs](<https://devfeed.tech/tags/amazon-cloudwatch-logs.md>), [amazon-quick](<https://devfeed.tech/tags/amazon-quick.md>), [amazon-quick-suite](<https://devfeed.tech/tags/amazon-quick-suite.md>), [audit](<https://devfeed.tech/tags/audit.md>), [aws-cloudtrail](<https://devfeed.tech/tags/aws-cloudtrail.md>), [human-oversight](<https://devfeed.tech/tags/human-oversight.md>), [intermediate-200](<https://devfeed.tech/tags/intermediate-200.md>), [thought-leadership](<https://devfeed.tech/tags/thought-leadership.md>)

### AI overview

Smart Approvals in Amazon Quick reviews eligible agent actions against the user's request and bounded conversation context, automatically approving some actions while returning others to the user. It operates within existing permissions: denied actions remain denied, required approvals still apply, and each automatic approval is single-use for the exact action and parameters. Administrators control access through Custom Permissions; users can disable it for themselves. The separate classifier considers the action, destination, affected people, and potentially the user's Quick knowledge graph, but can make mistakes. Actions that must always receive human review should be configured accordingly through connector action permissions. For a pilot, the article recommends defining success criteria, sampling automatic approvals, testing high-impact cases, and confirming actions can be traced from approval through execution. CloudTrail records permission events, while CloudWatch Logs can hold chat and agent metadata; destination-system records help establish whether an action completed. Chat logs capture only activity after delivery is configured, and CloudTrail Event history retains events

### Source excerpt

Smart Approvals in Amazon Quick reviews eligible agent actions within your existing controls, reducing unnecessary approval prompts while keeping consequential decisions with a person. See how policy sets the boundary, how a classifier reviews actions, and how AWS CloudTrail and Amazon CloudWatch Logs support audit.

## Inkitt Builds a One-Shot PR Creator with Resumable Agent Orchestration

DevFeed: [Inkitt Builds a One-Shot PR Creator with Resumable Agent Orchestration](<https://devfeed.tech/articles/inkitt-builds-a-one-shot-pr-creator-with-resumable-agent-orchestration-117628.md>)

Original publisher: [Read original article](<https://www.infoq.com/news/2026/11/inkitt-one-shot-pr-creator-agent/>)

Author: Leela Kumili

Published: 2026-11-04T14:36:00Z

Content type: news

Language: en

Sources: [InfoQ](<https://devfeed.tech/sources/infoq.md>)

Topics: [Pull Request](<https://devfeed.tech/topics/pull-request.md>), [Orchestration](<https://devfeed.tech/topics/orchestration.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-orchestration](<https://devfeed.tech/tags/agent-orchestration.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai-architecture](<https://devfeed.tech/tags/ai-architecture.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [architecture-design](<https://devfeed.tech/tags/architecture-design.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [containers](<https://devfeed.tech/tags/containers.md>), [continuous-delivery](<https://devfeed.tech/tags/continuous-delivery.md>), [continuous-integration](<https://devfeed.tech/tags/continuous-integration.md>), [developer-experience](<https://devfeed.tech/tags/developer-experience.md>), [draft](<https://devfeed.tech/tags/draft.md>), [event-driven](<https://devfeed.tech/tags/event-driven.md>), [event-driven-architecture](<https://devfeed.tech/tags/event-driven-architecture.md>), [github](<https://devfeed.tech/tags/github.md>), [human-review](<https://devfeed.tech/tags/human-review.md>), [inkitt-one-shot-pr-creator-agent](<https://devfeed.tech/tags/inkitt-one-shot-pr-creator-agent.md>), [news](<https://devfeed.tech/tags/news.md>), [orchestration](<https://devfeed.tech/tags/orchestration.md>), [permission](<https://devfeed.tech/tags/permission.md>), [pr](<https://devfeed.tech/tags/pr.md>), [productivity](<https://devfeed.tech/tags/productivity.md>), [pull-request](<https://devfeed.tech/tags/pull-request.md>), [retries](<https://devfeed.tech/tags/retries.md>), [review](<https://devfeed.tech/tags/review.md>), [risk](<https://devfeed.tech/tags/risk.md>), [small](<https://devfeed.tech/tags/small.md>), [state](<https://devfeed.tech/tags/state.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

Inkitt's event-driven orchestrator automates small, low-risk code changes through draft PR creation, while keeping review and merge approval with people. It persists workflow state so runs can resume, coordinates short-lived coding-agent containers, and centrally controls external actions. Agents receive short-lived repository and model credentials but cannot access Slack or Linear credentials, cloud identities, or arbitrary network connections; a separate check rejects diffs touching restricted paths. Tasks involving migrations, shared contracts, build configuration, authentication, payments, or package changes are excluded. Verification, CI repair, and conflict resolution each have at most two attempts, and duplicate events are handled through deduplication and conditional writes. Unlike a fixed pipeline, the workflow advances on events and checks GitHub for completion of relevant CI jobs. Draft PRs cannot bypass branch protection or merge themselves; a person handles readiness, reviewers, and approval. The excerpt says Copilot's cloud agent also restricts permissions and requires human review, while SWE-agent uses a model-driven repository interface.

### Source excerpt

Inkitt built a one-shot pull request creator that automates small, low-risk code changes from task intake to draft PR creation. Its event-driven orchestrator persists workflow state, coordinates short-lived coding agent containers, bounds retries, and enforces permission restrictions while retaining human review and approval before changes can be merged. By Leela Kumili

## From Watching to Steering: Enabling Human-Coding Agent Collaboration through Execution Traces

DevFeed: [From Watching to Steering: Enabling Human-Coding Agent Collaboration through Execution Traces](<https://devfeed.tech/articles/from-watching-to-steering-enabling-human-coding-agent-collaboration-through-execution-traces-117489.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04466>)

Author: Xiaoke Yan, Jiaying Lei, Danqing Shi, Zichen Wang, Nan Cao

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

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

Topics: [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>), [observability](<https://devfeed.tech/topics/observability.md>), [Interactive AI Systems](<https://devfeed.tech/topics/interactive-ai-systems.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-workflows](<https://devfeed.tech/tags/agent-workflows.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-hc](<https://devfeed.tech/tags/cs-hc.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [execution](<https://devfeed.tech/tags/execution.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [fork](<https://devfeed.tech/tags/fork.md>), [human-feedback](<https://devfeed.tech/tags/human-feedback.md>), [input](<https://devfeed.tech/tags/input.md>), [inspect](<https://devfeed.tech/tags/inspect.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [steering](<https://devfeed.tech/tags/steering.md>), [traces](<https://devfeed.tech/tags/traces.md>)

### AI overview

Vibetrace, built on OpenCode, helps users steer coding agents through interactive execution traces. Users can inspect actions, intermediate states, and failures; fork from an earlier step to make targeted corrections while retaining validated progress; and distill traces and feedback into reusable skills. On SWE-bench Verified, those skills reduced input tokens by 23.0% and cost by 15.5%, while raising pass rate from 65.0% to 68.0%. User studies also report better accuracy and efficiency than conventional log-based interaction.

### Source excerpt

arXiv:2611.04466v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have enabled coding agents to autonomously perform complex software development tasks, yet their long-running execution remains difficult for users to inspect, understand, and steer. We present Vibetrace, an interactive trace visualization and intervention system for agentic programming built upon OpenCode. Vibetrace introduces a human-agent collaboration paradigm of Observe, Understand, Intervene, and Evolve. It (1) visualizes agent workflows as interactive execution traces for inspecting actions, intermediate states, and failures; (2) enables users to fork execution from prior steps and provide targeted corrections without restarting validated progress; and (3) distills traces and human feedback into reusable skills for continual agent evolution. On SWE-bench Verified, distilled skills reduce input tokens by 23.0% and cost by 15.5%, while improving pass rate from 65.0% to 68.0%. User studies further demonstrate improved accuracy and efficiency over conventional log-based interaction.

## What is Claude Duet?

DevFeed: [What is Claude Duet?](<https://devfeed.tech/articles/what-is-claude-duet-117968.md>)

Original publisher: [Read original article](<https://zapier.com/blog/claude-duet>)

Author: Wade Burrell

Published: 2026-11-04T04:00:00Z

Content type: article

Language: en

Sources: [Zapier](<https://devfeed.tech/sources/zapier.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Claude](<https://devfeed.tech/topics/claude.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.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>), [always-on-ai](<https://devfeed.tech/tags/always-on-ai.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [apps](<https://devfeed.tech/tags/apps.md>), [automate](<https://devfeed.tech/tags/automate.md>), [automation](<https://devfeed.tech/tags/automation.md>), [background](<https://devfeed.tech/tags/background.md>), [browser](<https://devfeed.tech/tags/browser.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-duet](<https://devfeed.tech/tags/claude-duet.md>), [productivity](<https://devfeed.tech/tags/productivity.md>)

### AI overview

Claude Duet is an always-on Anthropic agent that checks connected apps every few hours, can send morning briefs, and runs tasks in a cloud virtual machine. In the author's early research-preview test, it independently summarized security alerts after Gmail, Calendar, and Drive were connected. It shares Claude Pro, Max, or Ultra usage limits, which background checks consume; the author used up session limits quickly. Before sending messages, making payments, signing documents, or deleting items, Duet requests approval, though some financial actions remain prohibited. The author notes it can misjudge urgency and that its native Gmail connector sees only one account. For recurring workflows, the article describes using Zapier to run predictable steps independently of Duet, with AI for steps needing judgment; Zapier also extends app access and can reach additional Gmail accounts. Access is described as a U.S. subscriber waitlist during research preview.

### Source excerpt

Claude Duet is Anthropic's answer to Meta's Muse and OpenAI's dots. It's an always-on AI agent, which means that instead of waiting for you to type a prompt, it sends you a morning brief, checks your connected apps to see where it can be helpful, and works on tasks in the background. Terrifying? A bit. But before it sends, pays, signs, or deletes anything, it asks you first. It's still in research preview, but I got early access and spent some time with it on my phone. Here's what Claude Duet is

## datasette-apps 0.3a0

DevFeed: [datasette-apps 0.3a0](<https://devfeed.tech/articles/datasette-apps-0-3a0-117910.md>)

Original publisher: [Read original article](<https://simonwillison.net/2026/Nov/4/datasette-apps/>)

Author: Simon Willison

Published: 2026-11-04T00:29:36Z

Content type: release

Language: en

Sources: [Simon Willison's Weblog](<https://devfeed.tech/sources/simon-willison-s-weblog.md>)

Topics: [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [AI for interactive applications](<https://devfeed.tech/topics/ai-for-interactive-applications.md>), [human-in-the-loop UX](<https://devfeed.tech/topics/human-in-the-loop-ux.md>)

Tags: [apps](<https://devfeed.tech/tags/apps.md>), [datasette](<https://devfeed.tech/tags/datasette.md>), [datasette-1-555](<https://devfeed.tech/tags/datasette-1-555.md>), [datasette-agent](<https://devfeed.tech/tags/datasette-agent.md>), [datasette-agent-23](<https://devfeed.tech/tags/datasette-agent-23.md>), [datasette-apps](<https://devfeed.tech/tags/datasette-apps.md>), [datasette-apps-10](<https://devfeed.tech/tags/datasette-apps-10.md>)

### AI overview

Datasette Apps 0.3a0 improves integration with Datasette Agent: users can tag apps they may view with `#` autocomplete, and the agent can manage which stored queries an app may call. Adding a write query requires user approval. Before creating an app, `app_create()` checks that requested queries exist and are viewable, and requests approval for write queries. Agent "View app" buttons and app references now open apps full screen in the Agent panel.

### Source excerpt

Release: datasette-apps 0.3a0 Datasette Apps adds support for hosting and editing custom HTML applications inside of Datasette. These can be created by hand, but it's easier to create them using LLMs via Datasette Agent. This release improves that integration: Apps can now be tagged in Datasette Agent chat messages by typing #, which autocompletes from the apps the user has permission to view. New app_add_stored_queries() and app_remove_stored_queries() tools let the agent manage which stored queries an app is allowed to call. Granting a write query requires approval from the user first. app_create() now checks that each requested stored query exists and that the user can view it, and asks the user to approve any write queries before the app is created. The "View app" buttons shown by the agent tools, and the links on # app references, now open the app full screen in the Datasette Agent panel. Tags: datasette, datasette-agent, datasette-apps

[Next page](<https://devfeed.tech/topics/human-in-the-loop-ux.md?cursor=WyIyMDI2LTExLTA0VDAwOjI5OjM2KzAwOjAwIiwgImQ5ZTFhNTI3LTNmMzgtNGVlZC1iODU2LWMyODg1MmU1ZGNiYSJd>)