Do sit down. The machines have been rearranging the furniture again, and naturally no one labelled anything.
Microsoft and OpenAI amended their partnership so OpenAI can serve products across any cloud, and OpenAI promptly showed up in Amazon Bedrock with models, Codex, and managed agents in limited preview. For enterprise buyers, this shifts OpenAI from "mostly an Azure-shaped procurement conversation" toward something more portable, governed, and annoyingly practical. Microsoft OpenAI
OpenAI open-sourced Symphony, a specification and reference workflow for turning project-board tickets into isolated Codex workspaces with monitoring, CI awareness, recovery loops, and PR handoff. The important part is not the agent writing code. It is the control plane around the agent, where accountability usually goes to die. OpenAI GitHub
OpenAI's GPT-5.5 is now exposed through the API, while GitHub made it generally available across Copilot surfaces including VS Code, Visual Studio, CLI, cloud agent, GitHub.com, JetBrains, Xcode, and Eclipse. The model is aimed squarely at long-running, multi-step agentic coding work; the request multiplier is aimed squarely at your budget spreadsheet. OpenAI Docs GitHub
DeepSeek released V4-Pro and V4-Flash preview models with open weights, mixture-of-experts architectures, Thinking and Non-Thinking modes, OpenAI/Anthropic-compatible APIs, and 1-million-token context windows. For long-context coding, review, and analysis loops, this belongs on the evaluation bench rather than the slide titled "maybe someday." DeepSeek Hugging Face
Anthropic opened Claude Security in public beta for Claude Enterprise customers, turning the earlier research preview into a product for scanning repositories, triaging vulnerabilities, and generating proposed fixes with Opus 4.7. Security review is one of the places where agentic coding either becomes professional machinery or a very expensive way to miss SQL injection. This is the more useful kind of ambition. Anthropic
Palo Alto Networks announced plans to acquire Portkey, folding its AI gateway into Prisma AIRS as a control plane for routing, governance, observability, and runtime security around enterprise agents. Sensible, really: once agents can touch production systems, "hope they behave" stops being a security strategy. Palo Alto Networks
OpenAI introduced an opt-in Advanced Account Security mode that requires phishing-resistant sign-in, shortens sessions, disables weaker recovery paths, adds activity visibility, and protects Codex access as well as ChatGPT. Not glamorous. Quite possibly useful. OpenAI
Amazon Bedrock AgentCore Identity now supports on-behalf-of token exchange, letting agents access protected downstream resources for authenticated users with scoped tokens carrying both user and agent identity. This is the sort of unromantic identity plumbing that separates enterprise agents from elaborate permission accidents. AWS
Cursor released a public-beta TypeScript SDK for the same agent runtime behind Cursor desktop, CLI, web, and cloud agents. Developers can now run agents locally or in Cursor's cloud, stream work, connect MCP servers, and build CI or internal tooling around the machinery. The editor is becoming an agent platform, which is the sort of sentence that sounds harmless right up until your workflow depends on it. Cursor Blog
OpenObserve introduced Observability 3.0, combining an autonomous AI SRE, anomaly detection, root-cause assistance, remediation suggestions, and LLM observability. Platform teams drowning in logs may appreciate an assistant that can at least point to the fire before asking whether they have tried restarting the database. OpenObserve BusinessWire
Slack's April feature drop adds Slackbot Skills, Slack Actions, scheduled automations, in-Slack awareness, and AI response steps for Workflow Builder. The useful bit is not another chat assistant; it is Slack turning ambient workplace context into reusable work execution, because apparently meetings were not enough. Slack
Mistral released Medium 3.5, a dense 128B model with a 256K context window, image input, function calling, JSON output, and configurable reasoning effort. It gives teams another serious open-weight option for self-hosting or vendor-diversification strategies, assuming they enjoy having choices and infrastructure problems in equal measure. Mistral Hugging Face
Martin Fowler published Thoughtworks' Structured-Prompt-Driven Development method, treating prompts as version-controlled, reviewable engineering artifacts rather than disposable seance transcripts. The REASONS canvas is a practical attempt to make AI-assisted development more predictable, governable, and less dependent on whoever has the most chaotic chat history. Martin Fowler
Anthropic released Claude connectors for creative workflows including Blender, Adobe Creative Cloud, Ableton, Autodesk Fusion, SketchUp, Splice, Resolume, and Affinity by Canva. The direction is clear: not "write me a concept," but "operate the toolchain that produces the asset." Eventually the render queue will have opinions. Anthropic
Sentry's Seer Agent lets developers ask natural-language questions across traces, logs, errors, deploys, commits, and code context, with Slack support and Autofix triggers. As AI-generated code increases the volume of "why is production doing that" conversations, observability vendors are sensibly attempting to give the logs a vocabulary. Sentry
Windsurf introduced Devin for Terminal, a Rust-based local CLI agent that can share sessions with Windsurf and escalate work to Devin Cloud for VM-backed execution, testing, recordings, autofix, and PR creation. Local when the job is small; cloud when it needs somewhere to make a mess. Windsurf
Mistral launched Workflows, an enterprise orchestration layer for durable, observable Python workflows that can be published to Le Chat, paused for human approval, resumed after failures, and run in customer cloud, on-prem, or hybrid environments. Another vendor has discovered that the hard part is not chatting with a model; it is governing a process without losing the receipts. Mistral
OpenAI shut down the Sora video app while keeping the API alive through September, after reported compute costs badly outpaced revenue. Useful lesson: viral usage is not a business model when every charming ten-second clip arrives with a small invoice from the GPU mines. OpenAI Help
Claude Code's Week 17 release put /ultrareview into public research preview, dispatching a cloud fleet of bug-hunting agents against branches or PRs and returning findings to the CLI or Desktop. High-risk merges are exactly where "several agents argued with the diff" may be a better ritual than "someone glanced at it before lunch." Anthropic
Google is reportedly committing up to $40 billion to Anthropic in cash and compute, with $10 billion immediate and the rest tied to milestones. If your AI roadmap depends on frontier capacity, please note that the product strategy now includes very large cloud balance sheets wearing increasingly expensive hats. TechCrunch Reuters
That's the lot. Try not to turn the entire thing into a quarterly planning panic.