Ask AI's Console Rebuild: Evidence Inline, a Plan You Can See
TestRelic Team
Ask AI used to be two things sharing a screen: a chat panel, and a composer that felt like a separate app bolted onto it. Evidence for a claim — a screenshot, a trace, the actual failure — usually meant clicking through to a different artifact panel to go find it. This week's rebuild collapses that into one console.
Evidence where the conversation is
Test evidence — screenshots, traces, failure detail — now renders inline in the chat, not only inside a separate artifact you have to open. If the assistant says a login test failed because of a redirect loop, the trace showing that redirect loop is right there in the same message, not a click away.
See the plan before it runs
The composer now shows a context readout and the agent's own plan before it acts — so instead of watching a spinner and hoping, you can see what it's about to do and where it's pulling context from. Thread telemetry sits behind the context chip for anyone who wants to see exactly what the assistant used to answer.
Artifacts, authored directly
Generated artifacts — slide decks in particular — are now authored by the model directly instead of round-tripping through a separate renderer, closing a gap where a validator could false-reject a perfectly good deck and leave it stuck. Decks render live in the conversation as the model builds them.
Get started
Open Ask AI from any repository, or ask it to change how your tests are organized — the same console now backs Modules's chat-driven regrouping, using the identical docked panel.