One desk for the full equity loop, with markets, research, AI and analysis together in one place and on one record.
The desk
Markets, research, AI and analysis are rooms on one desk rather than four products glued together. Collaborative agents start the work in every room, and Copilot, Scout and the morning huddle travel with you, so your context never resets.
Named watchlists, charts, statements and the earnings calendar sit together, with a Copilot scoped to each ticker so company work never leaves the quote. The tape you read at the open is the same book you work all day.
Your living notes, thesis pipelines, files and diligence are kept here. Copilot drafts, links and versions its work into the system of record, so a session becomes a claim you can kill later rather than a transcript you lose.
Collaborative agents start the workflow for you: Scout, the coverage analysts, a Copilot in each session and a facilitated morning. You approve their missions and close the huddle, and they do not close the day, or the trade, for you.
Supply-chain, geography and fundamental screens send names into the same process that runs from idea to monitoring, rather than leaving them on a separate analytics island.
AI · Thesis pipelines
Pipelines hold the equity lifecycle, from watch through active research to the published memo, so agent output turns into desk work. Claims are versioned and falsifiable, which makes drift visible after the print. The AI drafts, and you keep the record.
AI · Markets & tickers
Named watchlists, fundamentals and charts support your bottom-up work, and Copilot is scoped to the name you are studying, so its answers stay grounded in that company rather than in a generic model.
Underneath
This is the same stack that runs the live portal, with model-agnostic agents, tools you can inspect and a research record that outlives the session, so the equity loop can keep compounding.
The desk is model-agnostic by design, with models from Anthropic, OpenAI and xAI selected per session, frontier IDs included. Every agent turn carries a budget ceiling and a tool trail you can inspect, and T3 is only used when chosen explicitly, never picked automatically.
Statements, prices, consensus and the earnings calendar across domestic and international coverage, drawn from the same data the ticker page, Copilot's tools and Scout already use.
When it is configured, screens and enrichment can use the in-memory engine, but Copilot, notes, markets and the morning do not depend on it.
Notes, pipelines, theses, messages and the audit trail are held point-in-time, so a claim can be replayed after the chat is gone. It is institutional memory that compounds across the equity loop, not a transcript you lose.
