A look at the features landing in the next three months across build, evaluate, and operate.
This quarter
Here is a look at the features landing in the next three months, grouped by the three things every team does with agents: build them, evaluate them, and operate them. As always, dates are directional — but the direction is firm.
What is shipping
- Build: richer eval templates, more first-party connectors, and a faster local dev loop.
- Evaluate: reference-free scoring on by default, plus regression gates in CI.
- Operate: per-environment guardrail policies and budget alerts on token spend.
- Platform: finer-grained roles and a public audit-log export API.
How we prioritise
Most of this list came straight from customer traces and support threads — the features that unblock real production work rise to the top. If something you need is missing, tell us; the roadmap is a conversation, and the fastest way onto it is a concrete story about what breaks today.