AI Predictability, redesigned
We’ve redesigned AI Predictability from the ground up. The new model is built on Claude Sonnet, and it brings higher accuracy, more transparency, and a clearer picture of how your engineering performance is trending in real time.
Where the old version gave you a forecast and asked you to trust it, the new one shows its work. With this update, you can:
- Pick any insight or graph in the platform and click the AI button.
- See Expected vs. Actual AVG, so you know not just where a metric is headed but how far it’s diverging from plan.
- Review personalized recommendations for closing the gap to your target.
- Deploy a Target Agent that tracks progress automatically, so you’re not the one who has to remember to check.
That last point is the real shift. Predictions are only useful if someone acts on them, and a Target Agent means the tracking happens whether or not anyone opens the dashboard that week.
A more flexible way to organize work
We’re also rethinking how organizations structure data inside Waydev. From now on, Groups replace Projects as the default view. When you create a Group, you choose whether it functions as an Initiative, Workspace, or Project, depending on how your team actually operates.
A quarterly cross-team initiative, a standing platform workspace, and a single-repo project all need different framing, and forcing them into one “Project” model was the limitation this replaces. This also isn’t a one-time decision: you can edit an existing Project at any point and redefine it as a Group, Initiative, or Workspace, so your structure in Waydev can evolve as your organization’s does, without starting over.
A few more upgrades
- Persistent team filters. They now stay set across pages, instead of resetting every time you navigate.
- Dynamic group redefinition. Projects can be redefined into different group types at any time.
- GitHub Copilot metrics. New metrics from GitHub Universe are being integrated into the Waydev AI Adoption module, alongside Cursor, Claude Code, and Windsurf.
- Waydev MCP. In development to bring these insights directly into Claude, Cursor, and other MCP-compatible tools, so you can ask about your engineering data from inside the tools you’re already working in.
Schedule a demo to see AI Predictability 2.0 in action.
Several items above have moved since this was first published. The Waydev MCP is no longer in development. It shipped, and now connects Waydev’s engineering intelligence to:
GitHub Copilot metrics are fully integrated and have since had a further round of accuracy improvements.
The bigger change is that this feature set has since become part of a larger platform release that goes well beyond what’s described above:
- AI Code to Production — following AI-generated code from IDE acceptance through commit, merge, and deploy
- AI Token Usage — across Cursor, Claude Code, Copilot, and Windsurf
- AI Checkpoints — commit-level attribution for AI-assisted work
- Waydev Agent — ask questions about your engineering data directly, customizable with
SKILL.mdfiles
If you’re evaluating Waydev today, that newer release is the more current picture of what the platform does. This post is best read as the update that started that direction, not the current feature set on its own.
See it in action
Schedule a demo to walk through AI Predictability, Groups, and the full platform release since.
Request a demo
