From the archive
AI agents use language models and software tools to plan tasks, write code and carry out multi-step work. Follow the latest AI agent news, coding assistants and automation frameworks, with attention to reliability, permissions and human oversight. Read the source-linked reports to understand what each system can actually do, where it still fails and what changes for developers and businesses.
Lead story
Anthropic’s Claude Code 2.1.282 combines a broad set of session and reliability fixes with new settings for users and administrators. The practical headline: it addresses several ways continued or resumed coding sessions could lose earlier reasoning, resend messages in altered form or fail outright.
Read the report & discussion ↗Microsoft AI Watch
Microsoft says its internal AI challenge has changed: getting people to use the tools is no longer the main obstacle, while governing agents that can act on their behalf is becoming the harder job. That matters beyond Redmond because the company is offering a practical picture of what…
Cohere Watch
TD Bank is putting up to $25m into a three-year collaboration with Cohere, bringing the Canadian AI company’s engineers alongside TD’s Layer 6 team in Toronto. The practical point is that this is aimed at building and testing AI inside a major bank, including agents and enhanced workflows,…
Watch Desk
A new Hugging Face guide argues that developers should inspect how AI agents find, choose and use tools, not merely whether the final task succeeded. That matters because an agent can abandon one product after repeated errors, complete the job elsewhere and still receive a misleading green tick.
Watch Desk
Docker and the Cloud Native Computing Foundation are proposing an open specification that packages an AI agent, its tools and its requested permissions inside a portable container image. The practical promise is unusually clear: an agent’s authority could travel with it across conforming runtimes, rather than being rebuilt…
Microsoft AI Watch
Microsoft and Rockwell Automation say an AI maintenance copilot at Rockwell’s Singapore factory has cut machine downtime by 33% and maintenance costs by about 25%. The system turns decades of shop-floor knowledge into answers technicians can use while diagnosing faults, offering a rare glimpse of AI doing something…
AWS AI Watch
Tech Mahindra and AWS have launched an Agentic Process Transformation Centre of Excellence aimed at moving enterprise AI agents beyond pilots and into operational business processes. The first joint deployment, a collections tool used by UK financial-services outsourcer Target Group, is reported to have delivered anticipated efficiency gains…
AWS AI Watch
AWS has published a reference architecture for AI agents that query data held across separate company accounts without copying the underlying datasets into one central store. The design uses Amazon Bedrock AgentCore Gateway and the Model Context Protocol to give an agent one governed endpoint for discovering and…
Google DeepMind Watch
Google DeepMind has launched Gemini 3.8 Live with Live Avatar, an enterprise feature that lets AI agents listen, see and answer with a generated visual persona in near real time. The important shift is practical: this is not merely a chatbot with a profile picture, but a system…
AMD Watch
AMD has crossed a US$1 trillion market value as investors pile into its AI hardware story, but the milestone is also a test: can revenue, supply and market share justify the price? The chipmaker remains far smaller than Nvidia in AI accelerators, so this is a bet on…
Meta AI Watch
A security researcher says Meta’s Muse desktop AI client for macOS contained an unpatched flaw that let locally running software redirect voice-dictation traffic to an attacker-controlled server. Meta has since deployed a hotfix, according to InfoQ, but the episode shows how an AI agent’s broad permissions can turn…
Watch Desk
EPC Group has published TAR-8, a framework for testing whether Microsoft 365 and Azure environments are actually ready for Copilot, AI agents and approved external models. Its useful idea is refreshingly concrete: do not ask only who has an AI licence, ask what the system can find, change,…
Watch Desk
As AI takes on more programming work, one new essay argues that programming languages should stop treating human-friendly syntax as the main design target. The proposed shift is towards stronger guarantees and richer program databases, a provocative idea with practical consequences for the tools people use to build…
Watch Desk
Contrastive-LM has released CLM-8B, an open Apache-2.0 model that scores candidate actions and returns probabilities instead of writing text. The practical significance is its attempt to make agent decisions faster, cheaper and easier to run locally, with the model’s head weighing 75MB and designed to run on one…
Watch Desk
Gortex, a new developer tool, turns code repositories into a searchable knowledge graph for AI coding agents. It runs locally as a single static binary, which means the useful bit is not another cloud dashboard, but a way to give agents structural context without shipping the repository to…
Watch Desk
Geordie has launched Cost Intelligence, a monitoring tool that links enterprise AI spending to the agents, tools, models and workflows generating it. The useful idea is simple: an AI bill can reveal a broken agent even when every conventional service still appears to be working.
Microsoft AI Watch
Microsoft is previewing an Integrated Security Operations Centre in Defender, a security platform designed to let people and AI agents detect, investigate and disrupt attacks from one shared foundation. The important change is architectural: Microsoft is joining security signals, context and protective controls so agents can act across…
Watch Desk
Meta has unveiled $1,299 VR glasses and a handheld Muse Charm designed to let people speak to its AI agent without unlocking a phone or opening an app. The bigger story is the attempted merger of immersive hardware and an assistant that can act on a user’s behalf.
Meta AI Watch
Meta has announced VR glasses that weigh about 100 grams, use a pocket-sized compute puck and are due to launch in spring 2027 for $1,299.99. The important shift is not merely making virtual reality lighter: Meta is combining a high-resolution display, hand and eye controls, an AI assistant…
Watch Desk
A new essay argues that AI data agents are running into the same workplace problem as people: crucial knowledge is scattered, implicit and poorly documented. Its proposed answer is less exotic than a new database architecture, and rather more useful: improve the documentation.
