Frequently asked questions

Let's walk through some questions you might have about FFP One: what it covers, how far you can trust it, what happens to your data, and how plans and the trial work. Each answer links to the page with the detail.

What's covered

Which companies and filings does FFP One cover?

The annual reports of over 2,300 companies that file with the SEC (10-Ks, and 20-Fs for foreign companies), plus their recent 8-K filings. FFP One reads each filing into its research database before anyone asks, so a question about margins or risk factors starts from the filing itself.

Quarterly figures (10-Q) aren't in yet. They're on the way, along with IPO filings and companies outside the US; the homepage lists the rest.

Does it cover mutual funds and ETFs?

Yes: the holdings of over 5,300 US mutual funds and ETFs, from each fund's N-PORT filings. You can look through a fund to the companies it owns, and see its figures added up as if it were one company. Funds publish their holdings quarterly and with a delay, so what you see is usually last quarter's portfolio.

How current is the data?

Share prices are daily closing prices, not live quotes. A Research Report is dated: it's a snapshot of what FFP One had on the day it ran, and it isn't updated afterward.

Can I get the data into Excel?

Yes, and exports use no credits. Figures download as CSV or Excel files, and Research Reports, Memos and company and fund reports export to PDF, Word or Markdown.

Trust and limits

How is this different from giving a 10-K to a general AI assistant?

A general assistant reads the document you hand it, one question at a time, and for a single filing that may be all you need. FFP One has already read the filings of over 2,300 companies and over 5,300 funds into the FFP One Data Layer, with standard measures worked out the same way for every company. So you can compare companies, screen a sector or look inside a fund in one question, and the AI looks the figures up instead of recalling them.

Every chat answer also says what it drew on and, when it used FFP One's data, lists the filings behind it. Why FFP One makes the longer case, including against building your own.

Where do the numbers come from, and are they checked?

From the database, not the model's memory. Standard measures (margins, P/E, return on equity, the Piotroski, Altman Z and Beneish scores) are calculated there in advance, and the AI looks them up. When a question needs a calculation nobody stored, like growth over five years, the AI works it out from the figures it looked up, and Chat checks the numbers in its answer against that data before you see it.

What companies say is a second path with its own checks. A filing's text is checked against the original SEC document before it's summarized, and every dollar figure and percentage in a summary is checked against the section it came from. Where the numbers come from has the whole picture.

Can it be wrong?

Yes. The checks are software and AI, not a person, and they catch a lot but not everything: a figure can come from the right filing and still be the wrong period or the wrong comparison. That's why every answer shows where its figures came from, so you can check the ones a decision rests on.

Is it investment advice?

No. FFP One is a research tool, not an adviser. It doesn't know your clients' circumstances, and nothing it writes is investment advice: it does the reading, and the decision stays yours. The system is designed to not give you buy, sell, or hold advice on individual equities.

AI models and your data

Which AI models does FFP One use?

Chat and Memos run on Anthropic's Claude. For each Research Report you choose the model: Claude (the default), OpenAI's GPT, or DeepSeek or GLM, the China-based models, which use a fraction of the credits. A Research Report can also search the web for recent news and keeps it in its own marked section; Chat never searches the web.

What happens if I choose a China-based model?

Your question, and any chat context you bring into the run, goes to that company's own service for that run only. Nothing else you do in FFP One reaches it. How they handle it is set out in the DeepSeek privacy policy and the Z.ai (GLM) privacy policy.

One caveat: these models can refuse or slant questions the Chinese government treats as sensitive, so for a topic like that, choose Claude or GPT.

Is my data used to train AI models?

FFP One doesn't use your content to train AI models, and doesn't sell your data. What each model provider does on its side is governed by its own API terms; the privacy notice lists every provider that receives your data, and says how to have yours deleted.

Where is my data stored, and how is it kept safe?

In the US, with Supabase (on AWS) and Google Cloud. The database is copied every six hours, and each copy is kept for 30 days. Every request goes through FFP One's server, which checks it against your account, and card payments go through Stripe's own checkout page. The Security page has the rest, including how to report a problem.

Plans, the trial and cancelling

How does the free trial work?

You get 500 credits and 10 days to try everything. A card is required to start, but the trial is never charged and never turns into a paid plan by itself: when it ends, you choose a plan or stop there. It's one trial per person. 500 credits cover a few Research Reports on Claude, or a dozen or more on the China-based models.

What uses credits, and what's free?

Anything the AI writes uses credits: a chat answer about 2 to 7, a memo about 2 to 4, and a full Research Report on Claude roughly 100 to 300 (roughly 5 to 40 on the China-based models). Browsing company and fund analytics, screening, filing search and data exports use none. A plan buys a monthly pool of credits for your whole firm; the pricing page has the plans and what each one covers.

Can my colleagues use it too?

Yes. Every plan covers your whole firm, with no limit on users and no per-seat fees: everyone draws on the same pool of credits.

How do I pay?

Plans are paid by card, and credit packs by card or PayPal, through Stripe's checkout. Prices are in US dollars. Monthly plans have no minimum term; annual plans are paid up front for twelve months.

Can I cancel, and can I get a refund?

You can cancel any time in the app. It takes effect at the end of the period you've paid for: you keep access until then and aren't charged again, and the unused part of the period isn't refunded. If you're a consumer, you also have the legal right to withdraw within 14 days, set out in the withdrawal notice.

If something has gone wrong, email support@ffp.one and we'll look at it. The Terms set out the details, including the one exception for a consumer's annual plan.

Didn't find your question?

Email support@ffp.one. If it comes up often, it'll end up on this page.