SEC Filings in ChatGPT and Claude: Connectors vs a Research App
Free connectors now let your AI assistant query 10-Ks, 10-Qs and transcripts with citations. Here is how to use them, and where a monitoring agent differs.
You can get SEC filings into ChatGPT or Claude in three main ways: upload the filing yourself, connect a filings data source through an MCP connector or plugin such as Equibles, or let the assistant search the web. Connectors are the most practical for regular use because the assistant can retrieve the right filing and cite it. None of these options watches your holdings for new filings or tells you when something changes, which is the job of a monitoring agent.
The difference between the options is mainly about where the documents come from and whether you can check the answer.
What are the ways to use SEC filings with an AI assistant?
| Method | How it works | Strengths | Weaknesses |
|---|---|---|---|
| Upload | Download from EDGAR and attach to the chat | Exact document; no setup | Manual; long filings may be truncated; one company at a time |
| Filings connector (MCP or plugin) | Assistant queries a filings database | Retrieves the right filing; can cite passages; works across companies | Coverage and limits vary; still needs verification |
| Web search | Assistant searches and reads web pages | Fast for recent news | May cite secondary sources rather than the filing |
| Dedicated research tool | Product built around filings | Designed for sourcing and comparison | Another tool to adopt |
What is a filings connector?
MCP, the Model Context Protocol, lets assistants such as Claude and ChatGPT call outside tools. A filings connector gives the assistant tools to search and read SEC filings and related data.
One example is Equibles. Its page describes more than 100 tools with every figure cited to the source filing, covering SEC filings including 10-K, 10-Q, 8-K, 20-F and 6-K with full-text search back to 2000, earnings call transcripts, 13F holdings, insider trades and other datasets. It lists a free plan with a daily request allowance, offers one-click connection for Claude and ChatGPT, and says the project is open source and can be self-hosted. Its ChatGPT setup guide describes installing it from ChatGPT's plugin directory.
Other providers offer similar access with different data. TradingView's MCP server beta, for instance, lets Claude analyse fundamentals, news, filings and calendars for users on Essential or higher plans. Paid data providers such as Fiscal.ai also offer connectors; see Fiscal.ai MCP vs a connected research app.
How do you ask good questions about filings?
Whichever method you use, the prompt matters:
- Name the filing and period. "In the 10-Q for the quarter ended 30 June 2026..." avoids the assistant picking the wrong document.
- Ask for quotes and locations. "Quote the passage and give the section."
- Ask for comparisons explicitly. "Compare Item 1A in this year's 10-K with last year's; list added and removed risks."
- Forbid outside knowledge when you want answers grounded in the document: "If it is not in the filing, say so."
- Check units and basis. Thousands or millions; GAAP or adjusted.
For a full section-by-section walkthrough, see how to read a 10-K with AI.
Which filing answers which question?
Knowing where information lives makes your prompts sharper:
| Question | Where to look |
|---|---|
| How does the company make money? | 10-K Item 1, Business |
| What are the main risks, and what is new? | 10-K Item 1A, compared with the prior year; 10-Q risk factor updates |
| Why did results change? | MD&A: Item 7 in the 10-K, Item 2 in the 10-Q |
| What is the guidance? | Earnings release, usually an exhibit to an 8-K |
| Who are the major customers? | 10-K business section and revenue or segment notes |
| Did an executive leave or a major contract change? | 8-K |
| Are insiders buying or selling? | Form 4 |
| Who are the large institutional holders? | 13F filings by the institutions |
A common mistake is asking for guidance from a 10-Q. Many companies give guidance in the earnings release and on the call rather than in the quarterly report, so the assistant may find nothing, or worse, find an old figure. Point it at the 8-K exhibit instead. Similarly, for questions about who owns a company, remember that 13F filings are made by the institutions, not by the company itself, and are filed after each quarter ends, so they show positions as of a past date.
What can still go wrong?
Connectors reduce errors, they do not remove them.
- Wrong document. Similar company names, share classes or amended filings (10-K/A) can be confused.
- Wrong period. Year-to-date figures in a 10-Q can be mistaken for quarterly figures.
- Table parsing. Complex tables can be misread, especially in older filings.
- Synthesis. Correctly cited figures from different periods can be combined incorrectly.
- Missing exhibits. Important detail is sometimes in exhibits, such as material contracts attached to an 8-K.
The habit that catches most of these: click the citation for any figure that affects your decision. Our article on whether you can trust ChatGPT for stock research covers the wider failure modes.
What does a connector not do?
A connector makes an assistant better at answering questions about filings. It does not:
- Notice that one of your holdings has just filed an 8-K
- Compare the new 10-Q with the last one unless you ask
- Remember why you own each stock and check new filings against it
- Watch suppliers and customers whose filings affect your holdings
- Reach you when you are not in the chat
Those are monitoring jobs. You can do them yourself by subscribing to EDGAR's RSS feeds and running comparisons when filings arrive, or use an agent built for them. See how to track changes in SEC filings.
Connector, research app, or both?
| If you... | Consider |
|---|---|
| Enjoy hands-on research sessions in your assistant | A filings connector |
| Hold a few stocks and check them occasionally | Uploads or a free connector |
| Hold many individual stocks and fall behind on filings | A monitoring agent |
| Want ongoing thesis checks with sources | A monitoring agent, plus a connector for ad hoc questions |
Try Ecomerate free
Ecomerate is an autonomous AI research and portfolio-monitoring agent for DIY investors. It watches your holdings and theses, flags filing and earnings changes, and texts you evidence-backed alerts with sources in iMessage.
Get started freeWhere does Ecomerate fit?
Ecomerate is an autonomous AI research and portfolio-monitoring agent for DIY investors, free to start. It handles the monitoring side: it watches your holdings and investment theses, detects changes in SEC filings and earnings, maps supplier and customer dependencies through Chains, and sends evidence-backed alerts with sources by iMessage. You can keep using a filings connector in your assistant for exploratory questions.
This article is for general information and research education only. It is not investment, tax or legal advice, and it is not a recommendation to buy or sell any security. Check material facts against current primary sources before making a decision.
Frequently Asked Questions
Can ChatGPT read SEC filings?
Yes. You can upload filings, use web search, or install a filings connector such as Equibles, which its documentation says is available in ChatGPT's plugin directory.
Can Claude access SEC EDGAR?
Through MCP connectors, Claude can query filings databases. You can also upload filings directly.
Are filings connectors accurate?
They retrieve real documents and can cite them, which helps a great deal. The assistant can still misread tables or periods, so check citations for important figures.
Do filings connectors send alerts?
No. They answer questions when you ask. For alerts on new filings, use EDGAR's RSS feeds or a monitoring tool.
Is it better to upload a 10-K or use a connector?
For a single deep read, uploading the exact filing is simple and reliable. For regular questions across several companies, a connector saves time because the assistant can find the right filing itself. In both cases, ask for quotes and check the figures that matter to you.