# waiting Published articles for waiting. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Send contracts for e-signature from n8n, with TurboDocx, Slack approval and SharePoint filing DevFeed: [Send contracts for e-signature from n8n, with TurboDocx, Slack approval and SharePoint filing]() Original publisher: [Read original article]() Author: n8n team Published: 2026-11-08T14:07:15Z Content type: tutorial Language: en Sources: [n8n]() Topics: [n8n](), [Slack]() Tags: [2](), [3](), [account](), [activation](), [by-partners](), [ceo](), [contracts](), [days](), [deal](), [guest-post](), [it-s](), [n8n](), [node](), [post](), [send](), [sharepoint](), [slack](), [spotlight](), [this](), [waiting](), [with]() ### AI overview The workflow takes contracts from an Excel row marked "Ready to send" through template generation, Slack approval, e-signature, and SharePoint filing. n8n schedules and coordinates the steps; TurboDocx generates a Word-template agreement as a PDF, sends it for signature, and triggers a webhook when signed. Slack approval gates sending, while Excel tracks Sent, Rejected, and Signed status. Setup requires n8n, the TurboDocx node and account, Microsoft 365 credentials, and a Slack workspace. The example supports one signer and does not handle redlines or negotiation. For self-hosted n8n, the signature webhook requires a public HTTPS callback URL. ### Source excerpt This Verified Node Spotlight is a guest post written by Nicolas Fry, CEO of TurboDocx. The deal is closed and the contract is ready. It's been sitting in someone's inbox for three days, waiting for an "OK, send it." Once it's finally ## Marvell Cuts To The Front Of The Datacenter Optics Line With Lumentum Deal DevFeed: [Marvell Cuts To The Front Of The Datacenter Optics Line With Lumentum Deal]() Original publisher: [Read original article]() Author: Timothy Prickett Morgan Published: 2026-11-03T21:31:01Z Content type: opinion Language: en Sources: [The Next Platform: In-depth coverage of high end computing]() Topics: [systems](), [communications](), [AI Infrastructure](), [Finance]() Tags: [2](), [2015](), [2020](), [2022](), [2025](), [about](), [accelerators](), [acquisition](), [connect](), [cpo](), [datacenter](), [deal](), [decades](), [like](), [lumentum](), [marvell](), [npo](), [optical](), [optical-interconnects](), [optics](), [photonics](), [silicon](), [silicon-photonics](), [waiting](), [with]() ### AI overview The excerpt presents Marvell's planned $80 billion stock acquisition of Lumentum, plus assumed debt, as a way to secure scarce optical components for AI datacenter systems and strengthen Marvell's custom chip business. It says Lumentum's indium phosphide lasers are a key supply constraint: orders are reportedly booked through early 2029, and neither Lumentum nor rival Coherent alone can meet Nvidia's projected needs. Marvell could prioritize supply for its own customers while continuing to sell to other chip and AI companies. Combining their expertise may make vertical integration and volume supply more credible, though the article flags customer concentration risk. The deal also targets near-package and co-packaged optics as bandwidth demands move from datacenter links toward rack, system, and package connections. Its completion is expected in the first half of 2028, with possible regulatory challenges in China and Europe and a long window for counteroffers. ### Source excerpt For a long while, like several decades now, many of us have been waiting for silicon photonics to co ... ## Your AI Assistant Is Waiting Around - An AI Agent Goes to Work DevFeed: [Your AI Assistant Is Waiting Around - An AI Agent Goes to Work]() Original publisher: [Read original article]() Author: Elnaz Sarraf Published: 2026-11-02T22:09:36Z Content type: article Language: en Sources: [HackerNoon]() Topics: [AI Agent](), [cursor]() Tags: [2](), [3](), [about](), [abstract](), [account](), [actions](), [agent](), [agent-architecture](), [agentic](), [agents](), [ai](), [ai-agent](), [ai-agents](), [ai-agents-explained](), [ai-agents-tutorial](), [ai-assistant](), [algorithms](), [artificial-intelligence](), [automation](), [chatbot](), [company](), [future-of-work](), [how-to-build-an-ai-agent](), [instructions](), [takes](), [waiting](), [work]() ### AI overview The article defines an AI agent as a system given a goal and guardrails that can gather information, reason through steps, take actions, retain context and pursue an outcome. It contrasts a chatbot that recommends restaurants with an agent that researches, checks availability, compares prices, books and adds a reservation to a calendar. The author's examples distinguish a knowledge-base bot from one able to process refunds and update a CRM, and GitHub Copilot's line completion from Cursor's described ability to work across a codebase, run tests and iterate. A suggested no-code starter uses Make.com: schedule a daily run, fetch a Google News RSS feed over HTTP, parse XML, filter items from the past 24 hours, summarize titles and descriptions with an OpenAI or Claude API module, then email a digest. The author recommends checking execution history if filtering drops too much and starting with simple recurring tasks before adding complexity. ### Source excerpt A chatbot waits for instructions. An AI agent takes a goal and does the work. A founder who built and exited an AI company explains the difference. ## Idle Cycles Are Not Wasted Cycles DevFeed: [Idle Cycles Are Not Wasted Cycles]() Original publisher: [Read original article]() Author: Matheus Stolet, Simon Peter, Antoine Kaufmann Published: 2026-11-02T05:00:00Z Content type: opinion Language: en Sources: [arXiv: Networking]() Topics: [scheduling](), [NUMA]() Tags: [amd-epyc](), [compute](), [cs-ar](), [cs-ni](), [cs-os](), [power](), [scheduling](), [service](), [systems-design](), [waiting]() ### AI overview The paper argues that idle service cores can return capacity to productive work when active cores are constrained by a shared package power and thermal budget. Measurements on a recent AMD EPYC processor indicate that the value of reclaiming idle cores depends on waiting strategy, processor topology, and idle duration; reclamation may deliver less benefit than commonly assumed while adding scheduling complexity. The authors propose treating power budget, rather than core occupancy, as the central resource and making waiting policy a first-class design choice. ### Source excerpt arXiv:2607.16408v2 Announce Type: replace-cross Abstract: Idle network service cores are treated as wasted compute. This assumption motivates increasingly sophisticated mechanisms that reclaim idle cores at microsecond timescales. We argue that this view no longer matches modern server hardware. On contemporary multicore processors, active cores compete for a shared package-level power and thermal budget. Once that budget becomes the limiting resource, an idle core that waits efficiently returns compute capacity that hardware can redistribute to productive work. Measurements on a recent AMD EPYC processor show how waiting strategy, processor topology, and idle duration determine this trade-off. Our results suggest that reclaiming idle cores often yields less benefit than commonly assumed while introducing substantial scheduling complexity. We propose a budget-centric view of service-core systems in which power, rather than core occupancy, becomes the fundamental resource and waiting policy becomes a first-class systems design choice. ## What Does "Now" Mean to Your Computer-Use Agent? Stale Screenshots, Page Loads, and Which Check to Run Before a Click DevFeed: [What Does "Now" Mean to Your Computer-Use Agent? Stale Screenshots, Page Loads, and Which Check to Run Before a Click]() Original publisher: [Read original article]() Author: Vasundra Srinivasan Published: 2026-10-30T04:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [computer-use](), [screenshot](), [Machine learning]() Tags: [agent](), [clicks](), [computer](), [computer-use](), [cs-ai](), [layout](), [loading](), [model](), [network](), [open-source](), [pixel](), [position](), [screenshots](), [this](), [vision-language-model](), [waiting](), [web]() ### AI overview In the measured pages, stale clicks were driven mainly by page loading: 17% of clicks planned in the first second after a page opened became stale within 30 seconds, compared with 0.4% on fully loaded pages. Waiting about one second for network activity to stop removed most risk on those pages, though eight of ten widely used open-source agents' wait rules still left over 1% stale clicks. Booking and buying pages in the HTTP Archive crawl settled more slowly and shifted layout more often than .gov pages. When a pre-click check is needed, pixel comparison verified target position as accurately as a frontier vision-language model at one eighth the latency; a 4B model matched the frontier model at checking whether the previous action succeeded. ### Source excerpt arXiv:2610.39282v1 Announce Type: new Abstract: Computer-use agents plan on a screenshot for several seconds and then click at a position on the screen; if the page has changed in the meantime, the click lands on an unintended element. Prior work treats the planning interval as the source of this risk. Measuring planned clicks on 31 public web pages, we find that the risk lies in page loading: 0.4% of clicks planned on fully loaded pages became stale within 30 seconds, against 17% of clicks planned in the first second after a page opened. On these pages, waiting about one second for network activity to stop removes most of this risk, yet the wait rules of eight of ten widely used open-source agents leave more than 1% of clicks stale. Commercial pages keep loading for longer: in the HTTP Archive's public crawl, 15% of 65 desktop booking and buying home pages went 500 ms without network requests within 3 seconds of their HTML being parsed, against 63% of 3,028 .gov pages, and 52% shifted their layout within 6.5 seconds of a network-idle wait capped at 3 seconds, against 8% (61% against 26% among pages recorded in full). Where a check before the click is needed, as before irreversible actions and on pages that are still loading, a pixel comparison verifies the target's position as accurately as a frontier vision-language model at one eighth of the latency, and a 4B model matches the frontier model in verifying whether the previous action succeeded. We release a benchmark generator, a break-even formula for when a check pays, and three rules for agent developers: https://github.com/multimodal-ai-engineering/preflight-bench ## Rank-Aware CPU Reservation and In-Place Adjustment for MPI CFD on Kubernetes DevFeed: [Rank-Aware CPU Reservation and In-Place Adjustment for MPI CFD on Kubernetes]() Original publisher: [Read original article]() Author: Tianfang Xie Published: 2026-10-30T04:00:00Z Content type: article Language: en Sources: [arXiv: Distributed Computing]() Topics: [Distributed Systems & Parallel Computing](), [High-Performance Computing](), [Computational Fluid Dynamics (CFD)]() Tags: [adjustment](), [cfd](), [controller](), [cpu](), [cs-dc](), [job](), [kubernetes](), [mesh](), [rank](), [throughput](), [waiting]() ### AI overview The framework uses MPI profiles to reserve CPU according to rank-level demand and adjust requests while OpenFOAM runs on Kubernetes, without hard CPU caps or changes to OpenFOAM, MPI, or the scheduler. It cut per-job reservations from 16.0 to 5.9 vCPU with at most a 6% increase in single-job wall time; six simulations on twelve nodes achieved 4.09x sequential throughput. Proportional request shaping showed no consistent benefit, and oversubscribing nominal CPU capacity caused throughput to collapse with default busy-waiting, though yielding waits softened the drop. In one run, 120 in-place resizes caused no pod restarts, and the pipeline reached 2.5x sequential throughput. Separately, reallocating ranks from far-field to near-wall mesh zones reduced fixed-iteration time by 2.0-4.2x versus the original concentric allocation and outperformed balanced Scotch partitions; results varied by mesh and iteration horizon. ### Source excerpt arXiv:2603.22691v2 Announce Type: replace Abstract: Non-uniform meshes in computational fluid dynamics (CFD) can leave some Message Passing Interface (MPI) ranks waiting while others determine execution time. On Kubernetes, reserving a full CPU for every rank can then restrict the admission of further simulations. We investigate rank-aware CPU reservation and in-place request adjustment using OpenFOAM on three Kubernetes testbeds. The framework lowers reservations without hard CPU caps and uses online MPI profiles to adjust each rank's CPU request while the solver runs. We reduced the per-job CPU reservation from 16.0 to 5.9 vCPU, with single-job wall-clock time increasing by at most 6%. With these reservations on twelve nodes, six co-located simulations reached 4.09 times the throughput of sequential execution. Matched controls showed no consistent benefit from proportional request shaping. Beyond the nominal CPU capacity, throughput collapsed with default busy-waiting; yielding waits softened the collapse. In one six-simulation run, the online controller performed 120 in-place resizes of running solvers with zero pod restarts; the pipeline reached 2.5 times the throughput of sequential execution. In a separate comparison on 16 physical cores, reallocating ranks from the far field to the near-wall zone reduced fixed-iteration wall-clock time by factors of 2.0 to 4.2 relative to the original concentric allocation, depending on mesh and iteration horizon. The reallocation also outperformed balanced Scotch partitions. The implementation requires no changes to OpenFOAM, the MPI library, or the Kubernetes scheduler. ## IonQ | Protect Network Corridors Before Native PQC Arrives | IonQ DevFeed: [IonQ | Protect Network Corridors Before Native PQC Arrives | IonQ]() Original publisher: [Read original article]() Author: Simon Patkovic Published: 2026-10-29T16:41:37Z Content type: opinion Language: en Sources: [IonQ]() Topics: [Post-quantum cryptography]() Tags: [2](), [2026](), [3](), [accelerate](), [added](), [adversary](), [data](), [decrypt](), [exposed](), [external](), [harvest-now-decrypt-later](), [ionq](), [keys](), [learn](), [native](), [network](), [pqc](), [quantum](), [risk](), [waiting]() ### AI overview Long-lived traffic can remain exposed to harvest-now-decrypt-later collection while network upgrades await vendor support and deployment. The article recommends prioritizing a small number of high-risk corridors whose data must remain confidential longer than their migration timeline. Where existing encryptors accept external keys, a separate key-management layer can supply PQC-derived material--and, on suitable controlled-fiber links, QKD-derived material--while native PQC migration continues. This can shorten the period of exposure without replacing encryptors or changing the data path; the article says key delivery adds no latency or throughput change. External-key support varies by product and release, so interoperability must be checked against deployed equipment. A lost key feed can be configured to fail open to native key establishment or fail closed, with an alert. The article also describes combining key sources to diversify cryptographic assumptions, while noting that QKD relays remain trusted infrastructure. ### Source excerpt Waiting for native PQC leaves long-lived data exposed to harvest-now-decrypt-later. Learn how external quantum-safe keys protect high-risk corridors today. ## Can Waiting Be a Game Mechanic in a Collectible Artwork? DevFeed: [Can Waiting Be a Game Mechanic in a Collectible Artwork?]() Original publisher: [Read original article]() Author: Nathan Meir Published: 2026-10-29T07:32:35Z Content type: article Language: en Sources: [HackerNoon]() Topics: [interactive storytelling](), [browser-game](), [User interface design]() Tags: [ai](), [ai-assisted](), [artwork](), [base](), [blockchain](), [browser](), [delay](), [design](), [disclosure](), [essay](), [exhibition](), [experience](), [experiment](), [game](), [interactive](), [interactive-nft-art](), [interactive-storytelling](), [nft](), [nft-collectibles](), [nft-escape-room](), [nft-storytelling](), [optional](), [plato-the-unseen](), [transformation](), [waiting]() ### AI overview Plato / The Unseen uses a fixed 24-hour pause after its fifth puzzle gate to create space before an optional, irreversible transformation into EIDOS. The artist frames the pause as a design proposition, not proof that a timer creates reflection, and says participants should understand its place in the journey and what the transformation entails; keeping the earlier artwork is also a valid choice. The browser-based NFT escape room runs on Base and requires wallet transactions and gas, which the interface should explain clearly. Its AI-assisted mood film conveys atmosphere, not gameplay. The artist says participation and understanding remain untested, and the collector mint is still being prepared. ### Source excerpt An artist's essay on the 24-hour pause and optional irreversible transformation in Plato / The Unseen, an interactive NFT artwork on Base. ## What happens when vendors take the lead on agentic AI DevFeed: [What happens when vendors take the lead on agentic AI]() Original publisher: [Read original article]() Author: Maria Korolov Published: 2026-10-28T10:00:00Z Content type: article Language: en Sources: [CIO]() Topics: [Agentic AI](), [AI for customer service automation](), [Artificial Intelligence](), [human review]() Tags: [agentic](), [agentic-ai](), [ai-stack](), [allows](), [apis](), [artificial-intelligence](), [artificial-intelligence-budgeting-c-suite-careers-cio-data-management-data-quality-emerging-t](), [budgeting](), [building](), [c-suite](), [careers](), [cio](), [connect](), [contact-center](), [customers](), [data-management](), [data-quality](), [edition-us](), [emerging-technology](), [forrester](), [hotel](), [it-leadership](), [it-management](), [it-strategy](), [language-en](), [like](), [money](), [restaurants](), [revenue](), [roi](), [storytype-feature](), [tasks](), [user-experience](), [vendor-management](), [vendors-and-providers](), [virtual](), [waiting]() ### AI overview The examples show vendors can provide ready-made agentic AI for focused tasks, while deployment still requires governance, cost control and employee adoption. Fertitta Entertainment used PolyAI call agents that connect to APIs to handle reservations; it reports abandoned calls fell from 40-50% to under 10%, attrition dropped 50%, and automated reservations generate 3,000 bookings and $600,000 monthly. Fogo de Chão reports all calls are now answered and bookings rose 10%; it also uses agents for repetitive internal work with human review. AltaClaro combines Verbit's agent platform with Credal for model flexibility and sensitive-data protection. The article cautions that agents can expose information through existing sharing permissions, production token costs can rise, and training alone may not change habits. It recommends readiness around information governance, workflow integration, ongoing adoption support and flexibility as tools and pricing change. ### Source excerpt There's no point building your own agentic AI stack for a non-core function when a vendor already has a product ready to go that meets your needs. For non-tech companies in particular, building from scratch can take time, money, and expertise they don't have. Take for example, Fertitta Entertainment, with more than 600 Landry's restaurants and the Golden Nugget hotel and casino. After the pandemic, the company had problems adequately staffing its call centers, says its director of contact center operations Brian Jeppesen. "The agents didn't want to come back," he says. The virtual assistants didn't work because the user experience was bad, and with the incoming call volume, 40 to 50% of them were abandoned. "Everybody wants AI, but nobody wants BS," Jeppesen says. "I talked to my CFO. This is what we're abandoning. This is the potential revenue we're losing. I can't throw bodies at it anymore." Agentic AI has upped their game for the virtual assistants. Instead of only answering simple questions like check-in hours, it allows the company to connect APIs and carry out more complicated tasks, like making reservations. "My abandonments fell to under 10%," says Jeppesen. "From waiting 45 minutes to an hour, they can get their questions immediately answered, or immediately transferred and taken care of." And turnover went down. Call centers typically see 100% turnover, says Jeppesen. Not because everyone quits, but because the bottom 25% quit and need to be replaced over and over again. Now that bottom tier doesn't have to be rehired, while the people who stay are the most knowledgeable. "They give a better experience for my customers," he says. "And it reduced my attrition by 50%." Forrester did a total economic impact study for the specific vendor that Fertitta used, PolyAI, and the ROI came back at 391% over the course of three years, with a payback period of less than six months. Jeppesen says he reviewed his own numbers and found that the automated reservations syst ## Nebius cut rollout collection time by 66.9% -- here's what changed DevFeed: [Nebius cut rollout collection time by 66.9% -- here's what changed]() Original publisher: [Read original article]() Author: Mark Chillingworth Published: 2026-10-27T15:38:21Z Content type: news Language: en Sources: [The New Stack]() Topics: [post-training](), [llm-reasoning](), [LLM optimization](), [grpo]() Tags: [actions](), [agent](), [agentic](), [agentic-ai](), [agents](), [ai](), [ai-agent](), [ai-agents](), [ai-engineering](), [ai-infrastructure](), [can](), [collection](), [first](), [gpus](), [heres](), [nebius](), [need](), [post](), [powerful](), [sponsor-nebius](), [sponsored](), [sponsored-post](), [stack](), [time](), [training](), [waiting]() ### AI overview Nebius reports that changes to routing, sandbox execution, and batch assembly cut collection time for a fixed-weight, 256-trajectory Terminal-Bench-based batch by 66.9%; the measure excludes training computation and is not a full training-speed result. The article argues that rollout systems can bottleneck agentic RL, so teams should track time to a complete trainable batch and its straggler tail alongside throughput. It recommends managing policy staleness by measuring divergence and applying corrections or filtering, building reproducible environments with reliable verifiers, separating infrastructure failures from model failures, and scheduling GRPO attempts by complete groups. It also covers reducing sandbox waits, aligning trainer and rollout numerics, and choosing weight-publication methods based on hardware and freshness needs. Jin advises smaller teams to begin simply and add asynchronous complexity when measurements show it is needed. ### Source excerpt In AI agent training, you can pay for powerful GPUs and still spend much of your time waiting. Agents need The post Nebius cut rollout collection time by 66.9% -- here's what changed appeared first on The New Stack. ## WaitLens: Temporal VLM-Based Adaptive Waiting for Industrial GUI Testing Agents DevFeed: [WaitLens: Temporal VLM-Based Adaptive Waiting for Industrial GUI Testing Agents]() Original publisher: [Read original article]() Author: Shika Ikeura, Xuan Wang, Ziyi Kuang, Jianping Xu, Chao Kang, Yongxiang Hu, Xiaoyu Wang, Juxing Yuan, Yangfan Zhou Published: 2026-10-26T04:00:00Z Content type: article Language: en Sources: [arXiv: Software Engineering]() Topics: [ui-testing](), [gui-agents](), [Machine learning](), [Advanced AI workflows]() Tags: [adaptive](), [agents](), [cs-se](), [gui](), [industrial](), [temporal](), [testing](), [vlm](), [waiting]() ### AI overview WaitLens targets wasted delay from fixed waits in industrial GUI testing: a study found that 42.3% of sampled interaction time was spent waiting, and 62.9% of that waiting continued after the GUI had finished loading. The method fine-tunes a lightweight VLM on GUI loading trajectories labeled with waiting end times; temporal resampling creates varied training examples to capture different loading behavior. It reached 86.08% accuracy in deciding whether to keep waiting, versus 54.13% for the backbone VLM. On recordings from Meituan's industrial testing pipeline, it cut mean extra waiting after the interface was ready by 94.98% compared with a 20-second fixed wait. These results are reported for that evaluation; the abstract does not provide broader deployment details. ### Source excerpt arXiv:2610.32482v1 Announce Type: new Abstract: Improving the efficiency of GUI testing agents is important for industrial deployment. Beyond LLM inference, fixed waiting between consecutive interactions introduces substantial overhead. Our industrial study reveals that fixed waiting accounts for 42.3% of the time within the sampled interaction intervals, comparable to the 54.8% spent on model calls. Moreover, 62.9% of the waiting time is spent after the GUI has already finished loading. Although prior studies have proposed adaptive waiting strategies, their reliance on app-specific data or platform-specific information limits their applicability across diverse industrial environments. To this end, we propose WaitLens, which fine-tunes a lightweight VLM on GUI loading trajectories annotated with waiting end times. We introduce temporal resampling to construct diverse training examples, enabling the model to learn varying GUI loading behaviors from temporal observations. Experiments show that WaitLens achieves 86.08% accuracy in deciding whether to keep waiting, compared with 54.13% for the backbone VLM. Our evaluation using recordings from Meituan's industrial testing pipeline shows that WaitLens reduces mean extra waiting after the interface is ready by 94.98% compared with a 20-second fixed wait. ## AI jobs are waiting while GPUs sit idle. Kubernetes has a new plan. DevFeed: [AI jobs are waiting while GPUs sit idle. Kubernetes has a new plan.]() Original publisher: [Read original article]() Author: Nick Lucchesi Published: 2026-10-23T19:08:53Z Content type: news Language: en Sources: [Kubernetes Overview, News and Trends | The New Stack](), [The New Stack]() Topics: [AIOps for Kubernetes](), [LLM optimization](), [OpenTelemetry]() Tags: [2026](), [ai](), [ai-infrastructure](), [ai-jobs](), [cloud](), [cloud-native](), [cloud-native-ecosystem](), [ecosystem](), [first](), [gpus](), [hpe](), [jobs](), [kubecon](), [kubecon-cloudnativecon-na-2026](), [kubernetes](), [native](), [observability](), [plan](), [post](), [road-to-kubecon](), [sponsor-hpe](), [sponsored](), [sponsored-post](), [stack](), [tracking](), [waiting](), [welcome]() ### AI overview Kubernetes AI workloads can leave scarce GPUs idle when scheduling and autoscaling operate out of sync. Proposed responses include DRA-aware autoscaling, workload-aware scheduling, provisioning requests, and APIs for direct coordination between the scheduler and autoscaler. The excerpt also describes HPE's use of an open-source tool to collect pod-restart diagnostics and send them to Slack, and llm-d v1.0's focus on production inference; an IBM demonstration reported 11.5x lower inference cost and about 98% cache reuse versus an all-proprietary configuration. For operations, a benchmark of 50 injected faults found Claude Code diagnosed more faults within two minutes when given read-only MCP context than with shell access. Metrics may miss transient Redis/Valkey state and change history; eBPF tracing helps with some visibility, but engine-level collection is needed for evictions, ACL denials, and configuration changes. ### Source excerpt Welcome to another edition of Road to KubeCon! We're tracking movements in the Kubernetes and cloud-native ecosystem as we count The post AI jobs are waiting while GPUs sit idle. Kubernetes has a new plan. appeared first on The New Stack. ## Introducing Infinity: a runtime for extreme agent concurrency DevFeed: [Introducing Infinity: a runtime for extreme agent concurrency]() Original publisher: [Read original article]() Author: Shadaj Laddad Published: 2026-10-23T16:58:32Z Content type: release Language: en Sources: [AWS Open Source Blog]() Topics: [Agent Harness](), [Multi Agent Systems](), [agent to agent communication](), [async](), [AI Agent](), [scheduling]() Tags: [about](), [acme](), [agent](), [agent-harnesses](), [agents](), [announcements](), [build](), [coding](), [coding-agent](), [concurrency](), [minutes](), [next](), [pull-request](), [repository](), [request](), [run](), [runtime](), [time](), [tools](), [triage](), [waiting]() ### AI overview Infinity is an early Apache 2.0 research runtime that suspends agents at tool calls, saves their conversation histories, and resumes them when queued messages arrive. Each bounded slice runs one model completion and dispatches at most one tool call, so idle agents hold no process or task; a benchmark reports about 100 KB per idle agent and an estimate of 75,000 agents in 8 GB under its tested setup. Per-thread queues preserve message order while different threads run in parallel. Step mode supports serverless execution, but requires durable state and queues, at-least-once delivery with deduplication, bounded slices, and serialized processing per thread; the AWS reference uses SQS FIFO, Lambda, Aurora DSQL, and DynamoDB. Synthetic tool calls represent later tool events and parent-child reports within model conversation formats. Events default to separate child threads, or can be added directly to the subscribing thread when marked associative. Open questions include how models handle interleaved events, scheduling priorities, compaction across agent trees, and event authenticity. ### Source excerpt Agents spend much of their time waiting on tools. A coding agent may start a build that runs for twenty minutes, a triage agent may wait hours for the next pull request on a repository, and an assistant may wait until the next morning for an approval. Most agent harnesses run the agent loop as [...] ## Luma API for AI Agents DevFeed: [Luma API for AI Agents]() Original publisher: [Read original article]() Author: Chaitrali Kakde Published: 2026-10-23T13:00:00Z Content type: tutorial Language: en Sources: [Swytchcode]() Topics: [API](), [Langgraph](), [human in the loop agent](), [REST API](), [Slack]() Tags: [actions](), [agent](), [agents](), [ai](), [ai-agent](), [ai-agents](), [airtable](), [api](), [apis](), [engineering](), [generation](), [how-to](), [jobs](), [minutes](), [run](), [takes](), [tokens](), [video](), [video-generation](), [waiting](), [webhooks]() ### AI overview Luma video generations run asynchronously, so submit a job, persist its generation ID, and have a scheduler or durable worker poll and wake the agent when it reaches a terminal state. The excerpt describes `POST /v1/generations` and `GET /v1/generations/{generation_id}`; it reports no documented webhook or callback, so polling is the described completion mechanism. Treat nonterminal states and caller-side timeouts as pending or unknown, not as failures that justify another paid generation. Use a deterministic idempotency key based on the work and check stored jobs before resubmitting; poll with widening intervals and a timeout ceiling. The article also recommends keeping API credentials outside the agent context and using argument-level approval rules to cap expensive generations. It cautions that its webhook information reflects documentation reviewed at the time. ### Source excerpt Video generation takes minutes, there are no webhooks, and agents are terrible at waiting. How to run long jobs on the Luma API without burning tokens or paying twice. ## Readiness Is Not Curiosity: Why Your Agency's AI Program Is Waiting on People, Not Models DevFeed: [Readiness Is Not Curiosity: Why Your Agency's AI Program Is Waiting on People, Not Models]() Original publisher: [Read original article]() Author: Austin Boone Published: 2026-10-23T13:00:00Z Content type: opinion Language: en Sources: [Everpure]() Topics: [shadow AI](), [AI Chat](), [data](), [human review](), [Learning](), [ai-experiments](), [Web Accessibility (a11y)]() Tags: [2](), [ai](), [ai-and-machine-learning](), [but](), [can](), [change](), [curiosity](), [data](), [everpure](), [first](), [learn](), [models](), [people](), [perspectives](), [pilot](), [plan](), [program](), [public](), [public-sector](), [waiting](), [with](), [workforce]() ### AI overview The post argues that public sector AI programs commonly stall after a successful pilot because workforce practices, operating models, and trusted data do not scale with the technology. It recommends approved tools and clear rules, role-based learning, accountable owners, human review, and governed, accessible data. It outlines a 90-day approach: choose and baseline one mission workflow, build and test a small version with its users, then measure adoption, quality, time saved, risk, and confidence before deciding whether to stop, improve, or scale. The excerpt also presents Everpure's internal training practices and Data Stream as examples, and cites three agency or school IT improvements that returned time or strengthened trust before AI adoption. ### Source excerpt Readiness Is Not Curiosity: Why Your Agency's AI Program Is Waiting on People, Not Models by Everpure Blog The pilot worked, but the agency didn't change. Why public sector AI readiness is a workforce change with a data dependency, plus a 90-day plan to prove your agency can learn and move. The post Readiness Is Not Curiosity: Why Your Agency's AI Program Is Waiting on People, Not Models appeared first on Everpure Blog. ## When Should AI Chime In? Benchmarking Timely Contribution of Proactive Agents in Multi-Party Chat DevFeed: [When Should AI Chime In? Benchmarking Timely Contribution of Proactive Agents in Multi-Party Chat]() Original publisher: [Read original article]() Author: Ruotong Wang, Brian S. Yu, Xinyi Zhou, Joseph Chee Chang, Pao Siangliulue, Jonathan Bragg, Liwei Jiang, Amy X Zhang Published: 2026-10-23T04:00:00Z Content type: article Language: en Sources: [arXiv: Artificial Intelligence]() Topics: [AI interaction best practices](), [Artificial Intelligence](), [human review]() Tags: [agents](), [ai](), [artificial-intelligence](), [benchmark](), [benchmarking](), [chat](), [collaboration](), [conversations](), [cs-ai](), [live](), [models](), [proactive](), [reasoning](), [relevance](), [timing](), [waiting]() ### AI overview PITCH benchmarks when an AI should contribute a fixed message in unscripted group chats, using 1,091 human judgments from 89 participants across 35 conversations. In live decisions, models achieved 19.0% best-turn accuracy, down from 57.8% when given the full conversation history; lay humans deciding live reached 23.7%. Model placements were also premature more often than human placements (46.4% versus 39.7%). The authors report that 93% of model justifications cite relevance, while almost none consider waiting, pointing to a need for agents to reason explicitly about timing. These findings are based on 15 models across 28 configurations. ### Source excerpt arXiv:2610.25896v2 Announce Type: replace Abstract: Proactive agents can support group collaboration by surfacing useful information without waiting to be asked, but even relevant contributions can be disruptive if poorly timed. In this work, we isolate the timing question: given a fixed candidate contribution such as a clarification or relevant fact, when should an AI agent chime in? We introduce PITCH (Proactive Intervention Timing in CHat), to our knowledge the first benchmark with large-scale human judgments of proactive agent timing collected from participants in the same unscripted multi-party conversations they took part in. To collect high-quality judgments, we design a novel interface in which 2-4 people chat synchronously, author candidate AI contribution messages, place their own and others' contributions into the chat history, and then deliberate as a group about the best placement for each message. PITCH contains 384 candidate contributions and 1,091 judgments, each marking an ideal turn, an acceptable window, or that a contribution should not appear, from 89 participants engaged in 35 conversations. Evaluating 15 models across 28 configurations, we find that while models can place contributions well when given the entire conversation history, they systematically chime in too early when asked to decide live as the conversation unfolds: best-turn accuracy falls from 57.8% to 19.0%, with nearly half of placements occurring before every participant's acceptable window. In comparison, lay humans deciding live are more accurate (23.7% vs 19.0%), and fewer of their placements are premature (39.7% vs 46.4%). Model reasoning traces show that 93% of intervention justifications cite relevance, while almost none consider waiting. By providing a human-grounded benchmark, our work reveals gaps in agents' ability to integrate into group settings and points towards proactive agents that can reason explicitly about appropriate timing. ## 7 Tips to Optimize Linux systemd Boot Time DevFeed: [7 Tips to Optimize Linux systemd Boot Time]() Original publisher: [Read original article]() Author: Ravi Saive Published: 2026-10-14T07:05:15Z Content type: tutorial Language: en Sources: [Tecmint: Linux Howtos, Tutorials & Guides]() Topics: [systemd](), [Linux]() Tags: [boot-process](), [linux](), [server](), [systemd-analyze-blame](), [systemd-boot-time](), [systemd-unit-dependencies](), [waiting]() ### AI overview This tutorial explains how to diagnose and reduce Linux systemd boot delays. It covers measuring startup time, inspecting slow units and the critical chain, safely managing unnecessary services, addressing network wait and filesystem mount delays, and configuring service dependencies. It recommends validating changes and comparing boot measurements after rebooting. ### Source excerpt The post 7 Tips to Optimize Linux systemd Boot Time first appeared on Tecmint: Linux Howtos, Tutorials & Guides . A Linux server that takes a minute to boot may be spending much of that time waiting for one or The post 7 Tips to Optimize Linux systemd Boot Time first appeared on Tecmint: Linux Howtos, Tutorials & Guides. ## Gradually expanding permissions for AI coding agents using approval data DevFeed: [Gradually expanding permissions for AI coding agents using approval data]() Original publisher: [Read original article]() Author: Nick Kilpatrick Published: 2026-10-01T12:00:20Z Content type: article Language: en Sources: [Atomic Object]() Topics: [AI-assisted coding](), [agentic-coding](), [Claude Code](), [codex](), [human in the loop agent]() Tags: [agent](), [agentic](), [agentic-coding](), [agentic-development](), [ai-agent](), [ai-for-developers](), [claude](), [claude-code](), [cli](), [codex](), [coding](), [development](), [trust](), [waiting]() ### AI overview The author describes a Codex permission system that records approval requests, groups similar operations by intent, and reports frequently approved candidates for broader permissions. The report suggests changes for human review but does not modify configuration; the author advocates reviewing and testing each rule before adoption. ### Source excerpt If you're anything like me, enabling that "Approve for me" setting in your agentic coding tool of choice feels like a big step. Waiting for approval prompts isn't the best use of my time, but I'm still skeptical enough that I don't trust the agent to decide entirely for itself when it needs my input. [...] The post Get Into Agentic Development From the Shallow End appeared first on Atomic Spin. ## fsGroupChangePolicy: OnRootMismatch and the setgid Invariant That Makes It Sound DevFeed: [fsGroupChangePolicy: OnRootMismatch and the setgid Invariant That Makes It Sound]() Original publisher: [Read original article]() Author: Philipp Lehmann (philipp.lehmann@gruppe.ai) Published: 2026-09-11T07:00:00Z Content type: tutorial Language: en Sources: [Philipp Lehmann]() Topics: [container](), [mount](), [Security](), [Process]() Tags: [container](), [container-image](), [kubernetes](), [mount](), [network](), [performance](), [reliability](), [security](), [stateful](), [storage](), [waiting]() ### AI overview This article explains why a stateful pod can remain in ContainerCreating for fifteen minutes when fsGroup triggers a recursive chown across a large network-backed volume. It describes how fsGroupChangePolicy: OnRootMismatch can skip the recursive walk when the volume root group already matches, and discusses the setgid invariant that supports this behavior. ### Source excerpt Pod stuck in ContainerCreating from fsGroup recursive chown: fsGroupChangePolicy: OnRootMismatch, and why the setgid bit on the volume root makes it safe. ## Waiting for PostgreSQL 19 - SQL Property Graph Queries (SQL/PGQ) DevFeed: [Waiting for PostgreSQL 19 - SQL Property Graph Queries (SQL/PGQ)]() Original publisher: [Read original article]() Author: depesz Published: 2026-07-31T16:57:44Z Content type: opinion Language: en Sources: [select \* from depesz;]() Topics: [PostgreSQL](), [SQL](), [Graphs](), [function](), [pattern matching]() Tags: [function](), [graph](), [graphs](), [pattern-matching](), [pg19](), [pgq](), [postgresql](), [property](), [sql](), [uncategorized](), [waiting]() ### AI overview The article discusses PostgreSQL's SQL/PGQ implementation for property graph queries, including GRAPH_TABLE, graph pattern matching, and related DDL commands. It notes that the change was later rolled back and expresses the author's uncertainty about the practical benefits and usability of the syntax. ### Source excerpt Important update This change has been rolled back. Discussion can be found here. On 16th of March 2026, Peter Eisentraut committed patch: SQL Property Graph Queries (SQL/PGQ) Implementation of SQL property graph queries, according to SQL/PGQ standard (ISO/IEC 9075-16:2023). This adds: - GRAPH_TABLE table function for graph pattern matching - DDL commands ... Continue reading "Waiting for PostgreSQL 19 - SQL Property Graph Queries (SQL/PGQ)" ## Waiting for PostgreSQL 20 - Add min() and max() aggregate support for uuid. DevFeed: [Waiting for PostgreSQL 20 - Add min() and max() aggregate support for uuid.]() Original publisher: [Read original article]() Author: depesz Published: 2026-07-09T12:41:14Z Content type: article Language: en Sources: [select \* from depesz;]() Topics: [PostgreSQL](), [DateTime]() Tags: [aggregate](), [btree](), [max](), [min](), [order](), [pg20](), [postgresql](), [sort](), [uncategorized](), [uuid](), [uuid-extract-timestamp](), [waiting]() ### AI overview This article discusses a PostgreSQL patch adding min() and max() aggregate support for the uuid type. It explains that uuid is totally ordered through comparison operators and a btree operator class, and demonstrates the aggregates with UUID v7 and random UUIDs. ### Source excerpt On 1st of July 2026, Masahiko Sawada committed patch: Add min() and max() aggregate support for uuid. The uuid type already has a full set of comparison operators and a btree operator class, so it is totally ordered. min() and max() were the only common aggregates missing for it. Add the uuid_larger() and uuid_smaller() ... Continue reading "Waiting for PostgreSQL 20 - Add min() and max() aggregate support for uuid." ## Waiting for PostgreSQL 20 - Add backend-level lock statistics DevFeed: [Waiting for PostgreSQL 20 - Add backend-level lock statistics]() Original publisher: [Read original article]() Author: depesz Published: 2026-07-06T17:30:55Z Content type: article Language: en Sources: [select \* from depesz;]() Topics: [PostgreSQL](), [Back end](), [Statistics]() Tags: [backend](), [pg-stat-get-backend-lock](), [pg-stat-lock](), [pg20](), [postgresql](), [statistics](), [uncategorized](), [waiting]() ### AI overview A walkthrough of a PostgreSQL patch that adds per-backend lock statistics, including lock wait counts, wait times, and fast-path exceeded counts. The author demonstrates the feature with concurrent psql sessions and discusses its usefulness for diagnosing lock behavior. ### Source excerpt On 30th of June 2026, Michael Paquier committed patch: Add backend-level lock statistics This commit adds per-backend lock statistics, providing the same information as pg_stat_lock. It is now possible to retrieve those stats (lock wait counts, wait times, and fast-path exceeded count) on a per-backend basis. This data can be retrieved with a ... Continue reading "Waiting for PostgreSQL 20 - Add backend-level lock statistics" ## Timers, Timeouts, and the Art of Waiting in Temporal DevFeed: [Timers, Timeouts, and the Art of Waiting in Temporal]() Original publisher: [Read original article]() Author: Joshua Smith Published: 2026-06-30T00:00:00Z Content type: tutorial Language: en Sources: [Temporal]() Topics: [SDK](), [Code](), [Process]() Tags: [concurrent](), [crashes](), [deployment](), [developers](), [how-to](), [scaling](), [sdks](), [temporal-concepts](), [timeout](), [waiting]() ### AI overview This tutorial explains that Temporal timers provide durable business-logic waiting, while Activity and Workflow timeouts detect failures. It describes how timers survive Worker or process failures and can support large numbers of concurrent waits. ### Source excerpt Timers and timeouts look similar in Temporal but solve different problems. Learn how each one works and how to choose the right tool for durable waiting. ## Waiting for PostgreSQL 19 - Add CONCURRENTLY option to REPACK DevFeed: [Waiting for PostgreSQL 19 - Add CONCURRENTLY option to REPACK]() Original publisher: [Read original article]() Author: depesz Published: 2026-04-21T10:40:14Z Content type: opinion Language: en Sources: [select \* from depesz;]() Topics: [PostgreSQL]() Tags: [bloat](), [concurrently](), [maintenance](), [pg19](), [postgresql](), [repack](), [uncategorized](), [vacuum](), [waiting]() ### AI overview The article tests PostgreSQL 19's proposed CONCURRENTLY option for REPACK on a table with 200 million rows. The test found that the table shrank as expected, while inserts experienced a visible but limited slowdown during repacking. ### Source excerpt On 6th of April 2026, Álvaro Herrera committed patch: Add CONCURRENTLY option to REPACK When this flag is specified, REPACK no longer acquires access-exclusive lock while the new copy of the table is being created; instead, it creates the initial copy under share-update-exclusive lock only (same as vacuum, etc), and it follows an MVCC ... Continue reading "Waiting for PostgreSQL 19 - Add CONCURRENTLY option to REPACK" [Next page]()