- Clarifo MCP Coverage: US filings and figures (2015–2026), plus data for listed companies in Sweden and Finland (2020–2026).
- MCP (Model Context Protocol) is an open protocol that connects Clarifo's data directly to Claude.
- You set up the connection once in Claude's connector settings; after that you query the data in conversation.
- Every answer returns figures with their source — no memorised or invented numbers.
- Worked example in this guide: Apple's five-year profitability trend in one question, with real 10-K figures.
What MCP is, and why it matters for Clarifo
MCP, or Model Context Protocol, is an open protocol published by Anthropic that lets Claude access external data sources and tools during a conversation. In practice it is a standardised way to give a language model access to exactly the data a task requires — with no copy-pasting, no separate export.
Clarifo exposes its normalised financial statement data as MCP tools. The same data available in the platform's Workspace is therefore also available in Claude through natural-language queries. When you ask Claude to retrieve a company's figures, it calls Clarifo's tools in the background and returns an answer in which every number is traceable to the original financial statement.
This is where the difference lies: the AI does not guess numbers from memory — it retrieves them from a structured source. That distinction is what counts when financial statement analysis feeds a decision.
Connecting to Claude
You add Clarifo MCP to Claude once, from the connector settings. The steps are as follows:
Open Claude's settings
In Claude.ai, go to Settings → Connectors. The exact menu labels may vary by subscription type and app version.
Add a custom connector
Choose to add a custom connector and paste Clarifo's MCP address: https://mcp.clarifo.com/mcp
Authenticate to your Clarifo account
The connection asks you to sign in to your Clarifo account. The data available to you depends on your subscription.
Start querying in a conversation
Once the connection is established, Clarifo's tools are available. Write your question normally — Claude selects the right queries for you.
Your first query: Apple's profitability over five years
Let's take a concrete example. The goal is to see how Apple's profitability has developed over five fiscal years — a question an analyst would otherwise answer by opening the 10-K filings and reconciling the figures by hand.
Apple's revenue grew from $365.8 billion (FY2021) to $416.2 billion (FY2025) — about 3.3% annual growth. Profitability improved faster: operating margin rose from 29.8% to 32.0%.
| Fiscal year | Revenue | Operating income | Margin |
|---|---|---|---|
| FY2021 | 365.8 | 108.9 | 29.8% |
| FY2022 | 394.3 | 119.4 | 30.3% |
| FY2023 | 383.3 | 114.3 | 29.8% |
| FY2024 | 391.0 | 123.2 | 31.5% |
| FY2025 | 416.2 | 133.1 | 32.0% |
Net income tells a different story than operating income. FY2025 net income was $112.0bn, up 19.5% year over year — well ahead of operating income's 8.0% rise. The gap comes from a one-off tax charge that weighed on FY2024 net income (tied to the EU State Aid ruling). The normalised operating income shows that operating profitability improved steadily — the net income "jump" is largely a comparison-year distortion, not a turn in the business.
Source: Apple Inc. Form 10-K, FY2021–FY2025 · Clarifo MCP (get_company_financials). Figures in USD billion.
This is Clarifo's editorial core: normalised data tells you not just how big a surprise is, but where it comes from. The gap between operating income and net income exposes a one-off item that net income alone would have hidden.
How traceability works
Every figure Clarifo MCP returns comes from normalised data, not from the model's memory. The answer carries the fiscal period-end date and a reference to the original financial statement or 10-K filing. The figures in the example above are Apple's officially reported numbers, with fiscal years ending in late September (FY2025: 27 September 2025).
In practice, this means you can ask Claude to show the source for any figure and trace it back to the original filing. The AI supports the analysis — it does not invent numbers. The same reliability you get in Workspace is available directly in your AI workflow.
Where to go from here
A single company's trend is a starting point. The same connection covers the core tasks of the workflow:
Peer benchmarking
"Compare the operating margins of these five Nordic industrial companies" — a peer set normalised on the same definitions.
Segment analysis
"Break down revenue by segment for the last five years" — the company's own reporting segments, with sources.
Cash flow trends
"Show operating cash flow and capital expenditure" — free cash flow development over time.
Questions and answers
Try Clarifo in Claude
Connect Clarifo's normalised financial statement data to Claude and query listed companies in natural language — with sources.
