Risk Dimensions · Analyst Seat
How It Fits Together
The shape of the system from your seat: where the data lives, where you work, and what your best work becomes.
The Data Hub
Read-only, shared
- Sovereign & fiscal data (debt, deficit, rates, macro)
- Markets: prices, securities, returns
- News, economic calendar, hedge-fund filings
- You can query all of it — never change it
→you read
Your Workspace
Yours to build in
- Your tools: Jupyter · Colab
- Pull hub data into your own LLM / AI workbench (your Claude CLI, notebook agent) — reason over it, draft the analysis
- Your private space to create tables, models, charts
- Isolated — your work is yours alone
→review & promote
Published Products
What great work becomes
- The Global Scorecard
- The Sovereign Fiscal Monitor
- New analyses the whole team uses next
The idea: you pull from the same data hub the firm runs on — into a notebook, a spreadsheet, or your own LLM / AI workbench — build your analysis with whatever tool fits the task, and the strongest results get reviewed and promoted into the products. The data is the contract: it flows out to your workspace and your finished work flows back, so every project leaves the team better positioned for the next one.
Two workflows from one seat
One read-only credential does both. The only difference is whether you also have permission to add to shared datasets — and even then, you add, never edit or delete.
Read-only · every seat
Explore & build
Pull from the hub → build in your workspace or your own LLM / AI workbench → ship for review → your strongest work gets promoted into a product. You can read everything; you can never change the firm's data.
Contributor · granted
Contribute to shared datasets
Everything above, plus push your output back on the same credential: append your work (e.g. your featured economic indicators) to a shared table → the house adds its lens (Coach insight) → a weekly grade scores how the call held up → that feedback drives your next cycle. Append-only — add records, never edit or delete, so the shared record stays clean.