For AI builders, quants & research teams
Pro tierPut a tireless analyst on every name you cover
Give your agent the Valuein connector so it reads structured, cited figures instead of writing numbers from memory, and let a human approve what matters.
Agents research, score, and flag changes across your universe while you sleep — and every figure the tools return cites its filing.
The pain
One analyst can deeply cover maybe 30 names. The rest of the universe goes unwatched. General-purpose LLMs can't fill the gap — they hallucinate financials, and you can't put a made-up number in front of an IC.
“My agent made up a revenue number.”
A model asked for a figure will produce one, correct or not.
“I cannot ship an agent my compliance team will approve.”
No source trail means no sign-off.
“I scrape filings and the parsing breaks.”
Unstructured text is a fragile base for an agent.
The solution
- 1
Give the agent verified tools, not guesses
Connect mcp.valuein.biz/mcp to Claude, ChatGPT, Cursor, or your own agent. Every tool call returns SEC-filed, point-in-time data with lineage — so the agent cites, it doesn't invent.
- 2
Run a flagship workflow in one call
The equity_research_brief playbook chains 8 tools into a cited brief (3 depth modes); screen_and_shortlist generates ideas and hands each to the brief.
- 3
Let it run — with a human on the loop
Agents execute a saved {team + goal + tickers + schedule} on the managed lane; agents can also schedule their own follow-ups and set rules that re-run the work when a filing signal fires. High-impact actions stage for your approval and land in an audit ledger — human-on-the-loop by default.
Copy this
Real code and a ready prompt. Each block says what it needs before it runs.
APython SDK
No token needed: runs on the free sample
from valuein_sdk import ValueinClient with ValueinClient() as client: # no key: the free sample # Structured figures plus the filing each came from: context an agent can cite. fin = client.financials("AAPL", freq="FY", periods=2) print(fin.summary()) print(client.filing_links("AAPL", form_types=["10-K"], limit=1))Runs on the free sample with no token. Use it to feed your own agent a cited context.
pip install valuein-sdk
BPrompt for Claude or ChatGPT
Works without a token on the free sample; your token unlocks your plan
Use search_companies to resolve the company, then get_company_fundamentals for its latest results. Quote every figure exactly as returned, cite its fact_id with verify_fact_lineage, and say so plainly if a figure is unavailable. Never state a number you did not receive from a tool.Tools it uses: search_companies, get_company_fundamentals, verify_fact_lineage
CConnect your AI client
No token? Delete the headers block and it runs on the free sample
{ "mcpServers": { "valuein": { "url": "https://mcp.valuein.biz/mcp", "headers": { "Authorization": "Bearer YOUR_VALUEIN_TOKEN" } } }}What you get
You cover the whole universe with the headcount of one — and every output is auditable to a filing, safe to put in front of an IC.
- An agent that quotes figures it was given, with the filing attached.
- A fact_id you can resolve to the source.
- A human stays in the loop for anything that changes state.
Related use cases
Frequently asked
How does it avoid hallucinating numbers?
The agent calls deterministic, typed MCP tools that return SEC-filed values with lineage — it doesn't generate financials from its weights. verify_fact_lineage links any figure back to its filing.
Can agents run on their own?
Yes — with a human on the loop. Agents fire on a schedule on Valuein's managed lane (model allow-list, deterministic tools), or interactively with your own BYO-LLM key. Mutating or outward-facing actions stage for your approval before they execute.
Try it on your own ticker.
The free sample needs no token and no card. One token works across your AI agent, your code and the Workspace.