The workspace
Reactor.
Where every question lands.
One chat surface. Routed automatically to the right Engine, the right system, and the right Skill — with everything streamed back in line: charts, narrative, SQL, citations, and ready-to-execute follow-ups.
Inside the Reactor
One surface. Every Engine. Every system.
Skills rail
Featured / popular / recent skills with engine filters and template-variable chips. One click to launch a parameterized run.
System routing
Keyword + weight scoring with an LLM fallback. Ambiguous? You see the routing decision and confidence — no black box.
Streamed reactions
Tokens stream as the model generates. Charts animate as rows arrive. Annotations overlay in real time.
Inspectable SQL
Every Insights answer ships with the exact SQL. Inspect, edit, re-run. Or pin the verified version as a Skill.
Charts that pick themselves
Bars, lines, areas, scatter, maps, big numbers — chosen for the result shape. Click a series to drill through.
Tags + storylines
Curate any thread of messages into a storyline. Tag once, find again. Sharable, exportable, replayable.
Followups, validated
Suggested next questions are pre-validated SQL. Click a follow-up and skip nl-to-sql entirely.
Export everywhere
CSV, XLSX, PDF, slides. One click. Includes the chart, the citation, and the run trace.
ContextChip
Each message bubble shows skill + system + routing source — so any answer is reproducible from its provenance.
The cycle
Question to answer to action — without leaving the chat.
More in the workspace
The Reactor keeps growing.
Conversations that resume
A session sidebar lists past conversations and replays the whole history when you pick one, so a Monday question can continue on Friday.
Pin to Home
Pin any message to the Watchlist widget on your Home canvas. It deep-links straight back into the conversation.
Pin a chart to Radar
Any chart in an answer pins to a live dashboard, and a back-badge on the tile keeps the link to where it came from.
Grounded in your documents
Point a System at a Repository collection and the Reactor answers from your own documents, with citations back to the source.
Tags
Tag any message inline. Tenant admins rename, recolor and clean up tags in one place.
Its own capability
Invoking the Reactor and AI inference is a dedicated permission, so a role can read data without being able to spend on models.