The World Model now opens on every category together, and clicking a node grows its connections on the same globe while the view zooms in on it. Anyone tracing a product area to the tasks and sessions behind it sees the whole neighbourhood without leaving the globe.
When an author replies "Fixed in <commit>", Code Review now reads the commented file at the head of the pull request before it answers. A thread resolves on code the reviewer has read, so a resolved finding can be trusted.
Automations now live in Ask Ellie. When a question is worth asking every week, one click under the answer schedules it, and Ellie suggests it for you once you have asked the same thing three times in seven days.
The World Model has a side panel where you can ask Ellie about whatever you are looking at. As it searches, the globe lights up the nodes it visits, so you watch the answer being found.
The Model Router now serves plain API traffic as well as coding agents. Set the model to "auto" on an OpenAI or Anthropic shaped request and the router picks the cheapest model that can handle that one request.
agentplay replays your coding agent sessions as a map and a timeline: what the agent read, what it edited, what failed and what it shipped. It runs locally with one command, and it is open source under Apache 2.0.
Agent Insights now has a Session Explorer: pick a session and step through what the agent read, edited, spent and retried, in order. The map lights each branch as the agent reaches it, and the timeline lanes show errors, cost and context at every step.
Every org can now open the World Model: one globe of the issues that keep recurring across your repos. Pick a pattern and it explains itself in a few bullets, such as how often it fires and what fixed it.
Bring a key for each provider your organization already pays, and Model Router routes through it. When none of your keys serves the model it picks, the turn runs on your Entelligence credits. A model policy keeps blocked models out of every route.
Read the releaseEvery review now scores two things separately: the quality of the code, and how much of the system it can break. The Review Scorecard combines them into a merge confidence you can trace back to its cause.
Model Quality compares the models your team codes with on the pull requests they actually produced. Each rate shows the sample behind it, so a small sample can never pass for a verdict.
Read the releaseAgent Insights shows what your coding agents cost in dollars, project by project against each budget. Open a project and its spend is charted week by week against the budget, so an overrun shows up before the month ends.
Read the releaseGetting the CLI working is now one command and one click in the browser. Signing in wires its skills into the coding agents it finds, such as Claude Code and Cursor, and the CLI updates itself, so the whole team stays on the same version.
Every production pattern in Incident Knowledge can now become a Linear or Jira ticket, with a team and an owner, from its own row. The ticket then sits on that row beside Fix, which runs Fix Now and opens a fix pull request.
From that day on, every session, review and incident lands in the same index.
