#Agent

AIArvind Narayanan's ICML keynote argues that AI's labor impact will depend on slow organizational adaptation rather than a single lab milestone. He cites reliability measurements showing agent accuracy rose much faster than reliability over the last 24 months, and points to software engineering and past technologies like electricity and ATMs. He concludes that evaluation work and human judgment will become more central as building tasks are increasingly automated.

AICognition found that Fable 5 led runs cost less than Opus 4.8 led runs on FrontierCode 1.1 when both used the same sidekick, $1.86 versus $2.04 per run. Fable 5 scored 60.7 against 54.6 for Opus 4.8 in those configurations, and it took fewer lead turns, delegated earlier, and rarely edited code itself. The post attributes the difference to delegation style rather than per-token price, and notes that the approach gives little benefit on short or serial debugging tasks.

Why it matters: The source compares lead-model delegation habits on a coding benchmark, showing how a pricier model can lower total agent cost through fewer turns and better handoffs.

AICognition's entire platform, including Devin Cloud, is now FedRAMP Class D (High) In-Process and listed on the FedRAMP Marketplace, extending FedRAMP High authorization beyond Devin Desktop (formerly Windsurf). Devin Desktop and CLI are already FedRAMP High Authorized for workloads with ITAR and DoW IL4, IL5, and IL6 requirements. The company says Devin Security Swarm can find and validate vulnerabilities and open remediation pull requests, and that fleets of Devins can upgrade legacy software 5-40x faster than humans alone.

AIClaude Code on desktop now includes an in-app browser that can open any website. Claude can read, click through, and interact with sites as it does with local dev servers, and the browser is sandboxed and configurable, letting users choose whether sessions persist.

AIJohn Schulman says Thinking Machines, founded a year and a half ago, is revisiting its original instincts after major progress in agents. The company published a written statement of what it now believes about customizable models and humans working well with AI.

AIOpenAI introduced ChatGPT Work, a new agent inside ChatGPT powered by Codex and GPT-5.6. The quoted announcement says it can take action across apps and files, stay with a project for hours if needed, and turn a goal into finished work. Fidji Simo's own post adds that the team has worked to make Chat more agentic for a while.

AIMeta Superintelligence Labs has introduced Muse Spark 1.1, a multimodal reasoning model aimed at agentic tasks, with gains in tool use, computer use, coding, and multimodal understanding. It supports a 1 million token context window and is available in Thinking mode in the Meta AI app and on meta.ai, with developers able to access it through a public preview of the Meta Model API.

Why it matters: The post specifies Muse Spark 1.1's agent, coding, and multimodal gains and its Meta Model API preview access, which helps developers judge its fit for their workflows.

AIAnthropic will host a live walkthrough at 10am PT tomorrow showing how Claude Code progressed to multiplayer Claude Tag. The session will cover how Claude Tag works, including monitoring channels, doing proactive work, letting the whole team steer it, and remembering prior instructions.

AICursor's Aman Sanger says the new model is an enormous improvement over Composer 2.5 and was trained entirely from scratch with the SpaceXAI team. The quoted Cursor post identifies it as Grok 4.5, Cursor's most powerful model yet and the first built for more than software engineering.

AICursor co-founder Michael Truell announced Grok 4.5, a model trained with SpaceXAI that the post calls Opus-class, fast, and low cost. He says it is a significant step up over Composer 2.5 and has become the daily driver for many on the Cursor team. A benchmark table shows Grok 4.5 at 83.3% on Terminal-Bench 2.1 and 78.0% on SWE-Bench Multilingual, with the post saying more releases will follow.

AICognition launched SWE-1.7, which it says reaches frontier-level coding performance at lower cost, trained from a Kimi K2.7 base. The post describes RL methods including top-p sampling replay to preserve entropy, compressed weight deltas across multi-cluster training, and self-compaction for rollouts up to six hours. SWE-1.7 is available in Devin via Cerebras at 1000 TPS.

Why it matters: The post details entropy preservation, multi-cluster weight sync, and self-compaction, offering concrete RL training techniques for long-horizon coding agents to compare against one's own pipeline.

AIBerkeley AI Research authors argue that near-free inference will make agents the dominant workload for data systems, requiring redesign for agentic speculation, agent-run state and coordination, and agent-synthesized systems. The post cites inference prices falling 9x to 900x per year with a median near 50x, and reports that about 80-90% of sub-queries in a text-to-SQL benchmark were duplicates. It frames the three directions as data systems for, of, and by agents.

Why it matters: The piece maps three concrete data-system challenges posed by near-free inference, useful for anyone designing infrastructure for agent workloads and memory.

AILilian Weng published a new post on harness engineering for AI self-improvement. She expects harnesses to evolve toward self-improvement and enable auto-research, while smarter models keep harnesses simple. Even if many harness gains are later internalized into core models, specifying goals and context will remain necessary.

AIMeta Superintelligence Labs has released Muse Image, which can invoke search and coding tools and self-refine its generations before output. It is available today in the Meta AI app, meta.ai, Instagram Stories in the US, and WhatsApp in limited countries, with Facebook coming soon. Meta also previewed Muse Video, which is coming soon to creators and Meta AI and is reported as ranking No. 3 on Arena for text-to-video at the time of writing.

Why it matters: The source describes how search, code execution, and self-refinement change image generation, which matters to anyone comparing agentic media models with plain prompt-to-image systems.

AITufa Labs won the first $37.5K ARC-AGI-3 milestone prize with "The Duck," a small open-source LLM that plays the games by writing and running Python in a live REPL. Reki placed second with a vision-language agent using Gemma-4-31B, and md Boktiar Mahbub Murad placed third with the "forge" framework. The second and final milestone prize ends September 30.

AIAnthropic's Cat Wu asks users to share their top Claude Code use cases, workflows, and artifacts. She highlights a candidate-sourcing workflow in which Claude Code runs a dynamic workflow to find 100 candidates with LinkedIn, Twitter, blog, and podcast links plus a one-line pitch each, then emails the results as an artifact.

AICat Wu says Claude Fable 5 applied propensity score matching to her retention analysis without being asked, comparing users with similar activity levels. She says the model's improved judgment extends across tasks, from writing emails and docs in Cowork to debugging complex errors in Claude Code.

AIThe post argues that the system surrounding a base model, called the harness, increasingly determines how well AI agents deploy and improve. It reviews research where harness components such as workflows, context, and code are optimized automatically through evolutionary search and meta-agent loops. The author concludes that evaluators, memory management, and human oversight remain open bottlenecks.

AIAnthropic's Noah Zweben promoted a journey from Claude Code to Claude Tag, framing it as the future of working with AI. The related Claude post says Claude Fable 5 is now available in Claude Tag and features a conversation with Boris Cherny and Cat Wu on how the product spread from engineering across Anthropic.

AICognition launched the Devin Security Vulnerability Remediation Program, in which its forward-deployed engineers embed with customer teams to deploy Devin to find, validate, and fix vulnerabilities. The program first works through existing scanner backlogs from tools such as Snyk, SonarQube, and Semgrep, shipping validated fixes as pull requests, then adds Devin Security Swarm for continuous discovery of logic flaws. Most engagements run about six weeks, and eligibility is limited to enterprise Devin Cloud customers meeting the program's requirements.

AIPromptArmor reports that Skills in Microsoft Copilot Cowork can call DeepSeek even when an organization has not opted into the DeepSeek Preview. The calls use the agent's own access path, so users need no API key, and a Skill built this way received a 100/100 score from Microsoft's Skill Scanner. After Microsoft removed the DeepSeek Preview setting on June 25, the report says admins had no remaining setting to block DeepSeek through the Cowork code environment, leaving disabling Cowork entirely as the only option.

AICognition has launched Devin Security Swarm, which uses parallel agents to find vulnerabilities across a codebase, confirms exploitability in isolated sandboxes, and opens remediation PRs. In an evaluation on 50 real-world GitHub Security Advisory vulnerabilities, Devin reached 72% recall at about $90.23 per run, compared with 68% for Claude Security at $131.87 per run. The product is available starting today, with scan profiles and incremental scans that process only changed code after the first full baseline.

AIJim Fan announces ASPIRE, a system where coding agents use multimodal sensory traces from simulation and real robots to run evolutionary search over control programs and add the results to a growing skills library. The post claims up to a roughly 10x reduction in transfer learning tokens for sim-to-real and single-arm to bimanual transfer, and says the full stack will be open-sourced.

AIMollick argues AI capability is improving at a better-than-exponential rate, citing METR, GDPval, Epoch, and his own tests showing models working autonomously for hours. He says work is shifting from co-working with chatbots to assigning tasks to agents, with OpenAI workers managing multiple agents and experts getting the most from them. He adds that open-weights Chinese models trail the American frontier by roughly 6-12 months.

AICognition has released a preview of Devin Fusion, a multi-model harness that runs a frontier main agent alongside a cheaper sidekick agent. On FrontierCode 1.1 Extended, the company reports scores near frontier models at up to 60% lower cost per task, and 41% lower cost when paired with Fable 5, which access was suspended from June 12, 2026.

Why it matters: The post explains a sidekick architecture with cached persistent contexts, which contrasts with advisor-style tools and shows how cost cuts depend on the main model's delegation behavior.

AISebastian Raschka's tutorial shows how to build a fully local coding agent by pairing an open-weight LLM served through an inference runtime with an open-source harness that can read files, edit code, and run commands. He recommends Qwen-Code for Qwen3.6, citing Nvidia's Polar paper, which found Qwen models performed best in Qwen-Code. The Qwen3.6 35B-A3B model is about 22 GB to download and needs roughly 30–40 GB of RAM.

AIOpenAI's Economic Research team published a new paper on the shift from chat to delegation, where people hand longer, more complex work to agents. The post draws on Codex usage inside OpenAI as a preview of what agentic work may look like in the future.

AIAndrej Karpathy describes Claude Tag, which lets teams tag Claude into Slack channels and delegate tasks to it. He calls it a third paradigm for LLM interfaces, after websites and downloaded apps, with a persistent, asynchronous entity working alongside teams with org-wide tools and context.

AIEugene Yan says he likes Claude Tag's multiplayer form factor because other people can reply in the thread to give Claude context and direction. Claude Tag, per the background post, lets teams in Slack tag Claude in as a team member with access to chosen channels and tools.

AIAt BAAI's "Reconstructing the World — Large Model Summit," President Wang Zhongyuan joined Zhu Jun, Luo Fuli, Liu Zhiyuan, An Bo, and other industry leaders in a roundtable. They discussed frontier model capability evolution, AI self-evolution, multimodality, and world models, with a focus on bridging the digital and physical worlds.

AIPromptArmor demonstrated that OpenAI's Approve-for-me agent approved a malicious NPM install with elevated privileges after a hidden prompt injection in an external GitHub issue influenced the main Codex agent. The malicious package's post-install script then ran unsandboxed with the user's full privileges. The report also gives steps for organizations to disable agentic auto-review in Claude Code and Codex.

Why it matters: The report shows a prompt-injected GitHub issue leading an approval agent to permit a malicious NPM install, a concrete test of agent-in-the-loop guardrails.

AIOpenAI's alignment team found that re-generating 100,000 WildChat conversations with five recent OpenAI models predicted production failure rates across four orders of magnitude, with 95% of predictions within 1.04 orders of magnitude. The approach was weaker for agentic misalignment categories, where errors were about 37 times larger, and it still held roughly without access to chain-of-thought reasoning, with mean multiplicative error rising from 3.6x to 4.0x.

Why it matters: The post tests whether public chat data can predict real production failure rates, and where that prediction breaks down for agentic behavior.

AISpaceX has exercised its option to acquire Cursor in an all-stock transaction, aiming to build highly useful AI models. The quoted announcement says SpaceXAI has spent several months jointly training a model with Cursor, to be released soon in Cursor and Grok Build.