A single-family office builds its own research desk, and cancels the subscription that used to sit behind it.
For years, Meridian Research, the research arm of a single-family office, did what most independent investors do: it bought its equity coverage from bank and broker desks. The reports were competent. They were also, fundamentally, not Meridian's.
Sell-side research is written for someone else. It's framed around a house view, published on the bank's calendar, and priced as a fixed annual subscription whether or not it matches what the investor actually needs to know. There's a structural issue underneath that, too: a bank that maintains an information wall between research and advisory still has an M&A team somewhere in the building taking companies public, and that bank's desk is never going to call that stock overpriced.
Meridian wanted something different: research it could shape and trust.
The requirements
To replace the sell-side desk, Meridian needed intelligence that was:
Grounded and consistent across filings, transcripts, technical reports, and broker notes
Structured for AI workflows, not delivered as raw or fragmented data feeds
Instantly accessible inside Claude, with no engineering overhead
Fully citable, so every material claim could be traced back to a retrieved document
The solution: Bigdata.com as the evidence layer
Meridian connected Claude to RavenPack's Bigdata.com through MCP and used it as the foundation for a nine-phase research workflow. The workflow chains eight Bigdata.com tools and roughly 70 individual calls into a single initiation report, and captures every document ID at the moment of retrieval, so nothing has to be reconstructed after the fact.
Bigdata.com supplies four things Meridian previously had to buy separately:
Company and market tearsheets: price, ownership, positioning, and commodity sentiment in one structured call
Filings and transcript search: subtype-precise, entity-anchored, date-bounded evidence on any covered name
Expert interview access: primary-source operator commentary on operating conditions and supply chains
Private document retrieval: Meridian's own research corpus queried alongside public content, under one citation scheme
Put together, one connector replaces what would otherwise be a filings vendor, a transcript vendor, a news terminal, an expert network, and a private document index.
Inside one initiation report
The workflow is designed to run unattended from a short prompt. An analyst enters a six-field brief, and Claude runs the full nine-phase sequence in order:
Proprietary scan, anchor, operational, macro, variant perception, peers, risk, catalysts, value and publish
Each phase pulls from Bigdata.com, and the pipeline finishes with a cited PDF and a full source appendix. In a typical initiation report:
Metric | Result |
|---|---|
Workflow phases (phase 0 to 8) | 9 |
Bigdata.com tools chained | 8 |
Bigdata.com calls per report | ~70 |
Uncited claims in the published report | 0 |
The result
Meridian now publishes institutional-grade initiation reports across a 43-name coverage book, on its own schedule, against its own thesis. It has cancelled its bank research subscription and moved to a pay-as-you-go model.
Impact metric | Without | With Claude and Bigdata.com |
|---|---|---|
Research source | Bank / broker desks | Self-produced |
Coverage book | Whatever was published | 43 names |
Cost model | Annual subscription | Pay as you go |
Auditability | Trust the desk | Cited to document ID |
As Meridian's principal put it:
"I have been reading equity research for a long time. What I produce now is structured exactly the way I want it, and there is no sales desk behind the rating."
One connector, one citation scheme
Meridian's analyst now works entirely inside Claude. Every figure in a published report resolves back to a specific, retrieved Bigdata.com document, replacing borrowed conviction with a research process the firm owns end to end.
Together, Meridian Research and RavenPack replaced a bought research subscription with a self-produced research system: trusted data, auditable citations, and coverage shaped entirely by the investor's own thesis.
