Give your AI the filed accounts.
Ask Claude, ChatGPT or Cursor for a balance sheet and get one — read from registry and exchange filings on 400M+ private and 60,000 public companies. Nothing estimated, nothing to build.
Works wherever your team already works
MonetaIQ speaks the Model Context Protocol over streamable HTTP. If your assistant supports MCP, it supports MonetaIQ.
Claude
ChatGPT
Cursor
GitHub Copilot
Windsurf
Le ChatThree steps, one config block
Get your token
Copy the API token from your MonetaIQ account, or start on a plan to generate one.
Add the server
Paste the config into your client. The location differs by client — the shape does not.
Ask a question
Your assistant discovers all 16 tools automatically and picks the right one per question.
{
"mcpServers": {
"monetaiq": {
"url": "https://mcp.monetaiq.com/mcp",
"headers": {
"Authorization": "Token YOUR_API_KEY"
}
}
}
}
# Verify the connection
curl -X POST https://mcp.monetaiq.com/mcp \
-H "Authorization: Token YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","method":"tools/list","id":1}'16 tools, mapped to the MonetaIQ API
Every tool is a documented endpoint — same data, same coverage, same rate limits as your API plan. Nothing is generated by a model.
Financial statements4 tools
Company identity4 tools
Ownership and group structure4 tools
Enrichment and monitoring4 tools
What your team will actually ask
These run end to end in one message. The assistant chains the tools itself.
Credit and underwriting
Risk teams- Is this applicant's working capital trend safe for a £250k limit?
- Flag any customer in this list whose current ratio dropped below 1.0 last year.
- Pull the last four filings and tell me when accounts started arriving late.
- Generate a credit report and summarise the three things I should ask about.
Investment research
PE, VC and analysts- Find UK software companies with £5–20m turnover and EBITDA margin above 15%.
- Benchmark this target against five closest peers on ROCE and gearing.
- Who owns this company, and what else do those shareholders control?
- Chart revenue and net margin for these four competitors since 2019.
Supplier and third-party risk
Procurement and compliance- Verify this supplier is active, then check whether net assets are negative.
- Which of our top 50 vendors have directors in common?
- Watch these entities and tell me if status or shareholders change.
- Confirm the legal entity behind this domain before we onboard them.
Corporate finance
FP&A and strategy- How does our gross margin compare to the sector median in our SIC code?
- Build a comparables table for these six companies and export the ratios.
- Which competitors grew headcount while turnover fell?
- Summarise the financial position of every company on this acquisition list.
Three ways to get financials into an AI workflow
Only one of them survives an audit.
| Assistant browsing the web | MonetaIQ MCP | Building your own integration | |
|---|---|---|---|
| Source of the number | Whatever page ranked | Filed accounts and exchange filings | Your chosen source |
| Private company depth | Sparse and inconsistent | 400M+ entities | Depends on the vendor |
| Traceable to a filing | No | Yes, per field | If you build it |
| Same answer when re-run | Shifts with the results | Identical, tied to the filing | Identical |
| Cost of a schema change | — | Handled by us | Your sprint |
| Works across clients | Per vendor | Any MCP client | Rebuild per client |
Built for teams that get audited
Your token, your limits
The MCP server carries the same permissions, entitlements and rate limits as your API plan. It cannot reach data your contract does not cover.
Every call is logged
Tool calls appear in the same usage metrics as API requests, so you can reconcile consumption by team and by month.
Sourced, not inferred
Data comes from official registries, stock exchanges and regulatory filings. The assistant reads what was filed — it does not estimate it.
GDPR compliant
Processing follows the same terms as the rest of the MonetaIQ platform, with encryption in transit and strict access control.
Revoke in one place
Rotating or revoking the token cuts off every connected assistant at once. No per-client cleanup.
Full data rights
Enterprise agreements include the usage and redistribution rights your product and compliance teams need in writing.
Before you connect
What is MCP, in one sentence?
The Model Context Protocol is an open standard that lets an AI assistant call an external data source directly, so you get live financials in the conversation instead of pasting a spreadsheet into it.
Do I need an engineer to set this up?
No. You paste one config block into your AI client and add your existing API token. There is nothing to deploy, host or maintain. If your client stores its config in a file rather than a settings panel, someone comfortable editing JSON will find it quicker.
How is this different from just using the API?
Same endpoints, same token, same coverage. The difference is who does the integration work. With the API your engineers decide which call to make and when. With MCP the assistant reads the tool descriptions and decides, so an analyst who has never seen the documentation can still get a balance sheet. Build on the API when you are shipping a product; connect over MCP when people need answers.
Do MCP calls count against my quota?
Yes. A tool call is an API call. It draws on the same profile allowance and rate limit as a direct request and appears in the same usage metrics, so consumption stays reconcilable by team and by month.
Does connecting an assistant send our data to the model provider?
Your question and MonetaIQ's response both pass through your AI provider. Whatever retention and training terms you have agreed with them apply to that exchange. Check those against your own policy before connecting accounts that touch customer or deal data — this is the one thing worth clearing with your security team first.
Can the assistant make a number up?
The figures come back from the endpoint exactly as filed. What the assistant adds is the commentary around them, and commentary can be wrong in the way any analyst's first read can be wrong. Treat the table as the record and the summary as a starting point.
How current is the data?
Public company financials are refreshed quarterly from exchange and regulatory filings. Private company financials update as new accounts are filed, which means freshness follows each registry's own filing deadlines rather than a fixed schedule.
Which jurisdictions are covered?
Coverage follows registry disclosure rules, and those vary sharply — some countries publish full accounts for private companies, some publish abridged figures, some publish none. Country-by-country detail is on the data coverage page, and it is worth checking before you build a process around a specific market.
Can we store or redistribute what comes back?
Standard subscriptions cover internal use. Caching results in your own systems, embedding them in a product you sell, or passing them to clients requires a redistribution agreement — a separate commercial arrangement rather than a plan upgrade. Raise it with sales early if it is part of your plan.
Can we run this on a dedicated deployment?
Enterprise agreements cover dedicated endpoints, IP allowlisting, custom rate limits and bulk delivery alongside the MCP server. Bring your security requirements to the first conversation and they can be scoped into the contract.
Give your AI a balance sheet.
Paste one config block and your assistant can read filed accounts on 400M+ private and 60,000 public companies.