- Brand
- 13F.ai
- Category
- Pending
- Primary Subcategory
- Pending
Integration details
Description
13F.ai puts the public filings investors read into ChatGPT: stock trades by members of Congress (STOCK Act) and by the President and executive-branch officials, insider trades from SEC Forms 3, 4 and 5, the quarterly 13F reports of institutional fund managers, 13D and 13G stakes, IPO registrations, mergers and tender offers, private placements, lobbying disclosures, federal contract awards, short interest, FDA actions, clinical trials, company financials and more. Ask in plain English: "Which members of Congress traded NVDA in the last 90 days?", "What did Berkshire Hathaway buy last quarter?", "Which Tesla insiders sold stock recently?". ChatGPT looks the answer up in the filings, and every row comes with a link to the official source (House, Senate, SEC and other agencies). Each person signs in with their own 13F.ai account, so ChatGPT only reads what that account can access. Every tool is read-only: nothing is written, traded or sent. Amounts in congressional trades are disclosed ranges, not exact figures, and filings can arrive weeks after a trade. This is information, not investment advice.
- Integration type
- Plugin
- Verification status
- Not applicable
- Platform
- ChatGPT
- Category
- Pending
- Primary Subcategory
- Pending
- Secondary Subcategories
- None listed
- Brand
- 13F.ai
- Access
- Account required
- First tracked
- 2026-11-13
- Tool count
- 42
- Market
- US
The broad Category that contains the Primary Subcategory.
The Primary Subcategory used for this profile’s headline score.
Other Subcategories where the Integration is listed.
Not discoverable in ChatGPT yet
ChatGPT now suggests Plugins on its own when they match a user's request, but not every Plugin is being surfaced yet.We haven’t seen 13F.ai suggested yet. Sign up and we’ll track it, and your Discoverability Score will appear here once ChatGPT starts surfacing it.
What discovery looks like

Get alerts for 13F.ai
Get updates when 13F.ai’s Discoverability Score or category rank changes.
Competitive lineup
42 tools agents can invoke
The average star rating and number of ratings of the main apps of listed companies on Apple's US App Store (about a hundred apps we chose, such as Uber, Netflix and Chipotle), one snapshot a day, and how each has changed. Collection started on 2026-10-23, so the history is short; the note says how long. Each row compares the app's newest snapshot with its oldest in the window (the last 30 days unless you give dates or a ticker): count_change and count_change_pct are the change in the number of ratings, average_change the change in the average. The average covers every version of the app since it launched, so it moves slowly; the count is the better sign of change, and it can fall when Apple removes ratings. Ratings are a rough sign of how many people use an app, not a measure of a company's sales. Google Play is not covered. source_url is the app's store page. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_app_ratings
Bills and resolutions in the House and Senate from Congress.gov, latest action first, one row per measure: the citation (bill_label, e.g. 'H.R. 1234'), title, sponsor with party and state, policy area, introduced date, latest action with its date, number of cosponsors, the Congress it belongs to and a link to the bill on congress.gov. Returns a page of rows plus total counts of matching measures. Filter by sponsor (member: Bioguide id or name), chamber it was introduced in, policy area, words in the title (keyword), Congress number (e.g. 119) and dates (since / until apply to the latest action). Introducing a bill is not the same as passing it: most bills never get a vote. Policy areas are assigned after introduction, so the newest bills may have none. Each row carries its official link (source_url). Free for every account.
get_bills
What a candidate's campaign raised and spent in a two-year election cycle, from the FEC's candidate summary files (2016 on): total receipts, total disbursements, cash on hand at the start of the cycle and at the last report, debts, the candidate's own loans, and receipts from individuals, from PACs and from party committees, with the date of the last report included (coverage_end_date) and a link to the candidate's FEC page. Newest cycle first, biggest fundraiser first within a cycle. member (a Bioguide id or a name) selects one member of Congress across cycles; or filter by candidate_id, cycle (e.g. 2026), office (house, senate, president), state, party (FEC code such as DEM or REP), status (incumbent, challenger, open_seat), members_only and min_receipts. With cycle and office and no member, the first rows are the top fundraisers. Amounts are for the whole cycle up to coverage_end_date, not a single election, and receipts include transfers between a candidate's own committees. Totals only: no individual donor is named or available (federal law bars using contributors' names commercially); what employees of a company gave in total is shown as employee giving by party on the company's page on the website. A donation to a campaign is not a vote and says nothing about intent; FEC reports arrive monthly or quarterly, so the newest figures lag. Free for every account.
get_campaign_finance
Industry-sponsored clinical studies registered on ClinicalTrials.gov, with phase, overall status, conditions, interventions (names and types), enrollment and the start, primary completion and completion dates. Views: completed_phase3 (Phase 3 studies marked completed, by completion date; the default with no company), updates (by last update posted; the default for a company), status_changes (changes of overall status that we observed, newest first; our history starts when we first stored each study, so it is not the registry's own history), upcoming (studies still open, soonest primary completion first). With no company only studies whose lead sponsor is matched to a public company are listed, from the last 90 days, unless you set listed to false or give dates. cik and ticker are set only for a confident name match (sponsor_match_method says how); a study with no ticker may still belong to a public company, and role any adds studies where the company is a collaborator. This is a register of studies, not results or a forecast: completed means the sponsor marked the study finished, not that the treatment worked; dates are the sponsor's own and an estimated date can pass without an update; terminated and withdrawn studies carry the sponsor's stated reason. source_url is the study's page on ClinicalTrials.gov. Information, not investment advice. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_clinical_trials
A company's financial statement figures from the XBRL data of its SEC filings (mostly 10-K and 10-Q): for the latest 8 quarters and 2 fiscal years (quarters and years change that), revenue, gross profit, operating income, net income, diluted EPS (USD per share), operating cash flow, cash and equivalents, long-term debt (the noncurrent part only, not total debt) and shares outstanding, all in US dollars unless said otherwise, and the newest revenue split by business segment, geography and product (members are the company's own names, not normalised across companies, and need not add up to total revenue). Takes a ticker or a cik. Each period lists its figures and the filings they came from (sources: form_type, accession_no, filing_url, and which figures each supplied); each source has a cite field holding its link in one sentence. The latest filing of a period wins, so a restatement replaces the figure and some annual figures come from another form, such as a proxy statement (DEF 14A, often rounded): form_type says which. Concept fallbacks per period: revenue is Revenues, then RevenueFromContractWithCustomerExcludingAssessedTax; diluted EPS falls back to basic EPS (fallback_concepts says when); shares outstanding is the balance sheet count, then the cover page count. Gross profit is revenue minus cost of revenue when the company tags none (derived lists it). Periods are the SEC's calendar-aligned frames: CY2025Q3 is the quarter that ends closest to the end of the third calendar quarter of 2025, whatever the company calls it; fiscal_year and fiscal_quarter are the company's own calendar, named for the calendar year in which the fiscal year ends. Fourth-quarter revenue, gross profit, operating income and net income are derived and flagged: the 10-K reports the whole year, so the quarter is the annual value minus the first three quarters of that fiscal year, only when all four exist for the same concept, the year is from a 10-K and the quarters are from 10-Qs (derived lists them, derived_from names the accessions used; a missing quarter leaves the figure out, and derived_skipped says when the annual figure is from a proxy statement). Fourth-quarter EPS is never derived. Operating cash flow exists for a first quarter and for a year only: 10-Q cash flow statements are year to date, those figures are not stored, so quarterly cash flow is not derived. Funds, trusts and companies that file 20-F or 40-F have few or no figures, and banks and insurers often lack the revenue, gross profit, operating income and long-term debt concepts. The revenue split's period_end is rounded to a month end and can lag the newest quarter by one quarter. Information from public filings, not investment advice. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_financials
One-call briefing on a public company: profile (sector, estimated market cap (market_cap, our estimate: shares from the latest SEC filing x last Tiingo close; market_cap_basis estimated, with market_cap_reason when it is null), shares with shares_as_of, and fund_ownership_pct: the share of the ticker's own class that 13F funds hold at the newest complete quarter; null, with fund_ownership_status implausible, when it would pass 150%), recent open-market insider trades with 12-month buy/sell totals, trades by members of Congress, institutional (13F) ownership with quarter-over-quarter changes, recent lobbying spend, federal contract awards and the latest new-data events. Each section has totals and source links. Sections that fail report an error without sinking the rest. For a Basic account the answer is the company's name and identifiers plus the congress trades section; the other sections are replaced by an upgrade message (not_included). The congressional part is free for every account; the other datasets need the Premium plan and a Basic account gets an upgrade message in their place. Rows carry their official source link.
get_company_overview
Compares one institutional manager's Form 13F positions in two of the quarters we hold (about six, any two): how many positions are new, added to, trimmed, exited or unchanged, and the positions themselves with the change in shares and in reported value, largest change in value first by default. manager_cik is the manager's CIK; period is the later quarter and prior_period the earlier one, by default the newest quarter and the one before it. Filter with change (new, added, trimmed, exited, unchanged), q (name or ticker) and position_type=options for puts and calls (their value is the notional amount of the shares they cover). Shares of the earlier quarter are adjusted for stock splits in between. A change in shares is not always a decision (index tracking, mergers and share-class changes move positions too) and value changes include price moves. This is information from public filings, not investment advice. 13F reports are filed up to 45 days after the quarter ends and list long US-listed positions only (no short sales or cash). This is information from public filings, not investment advice. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
compare_fund_holdings
Gifts that company political action committees (PACs) made to candidates for Congress, from FEC reports, newest first: the company behind the PAC (ticker and CIK when we matched it to an SEC filer), the PAC, the recipient (a member of Congress with party and state, or the candidate's committee name), the amount in US dollars, the date the PAC reports for the gift, the election it is for, the two-year cycle and a link to the FEC page. Returns a page of rows plus the number of matching gifts and their total dollars (counted over the newest 10,000 matches at most). Filter by company (ticker or cik), recipient (member: Bioguide id or name), committee, party, chamber, cycle (e.g. 2026), dates and minimum amount. A WHOLE COMMITTEE OR CHAMBER IS ANSWERED IN ONE QUERY: committee takes a committee code or name (e.g. 'House Financial Services', HSBA, 'Senate Banking') and covers its current members together, and group_by 'pac' returns the top PACs (company, total dollars, number of gifts, number of recipients) while group_by 'recipient' returns the top recipients (total dollars, gifts, number of PACs), each with an FEC link. Example: top PACs giving to the House Financial Services Committee this cycle = committee 'HSBA', group_by 'pac'. By default only PACs whose company we matched to an SEC filer are listed (set include_unmatched true for every PAC), except with committee, where every corporate PAC giving to the committee's members is counted (private companies and partnerships too; each PAC group has a company_label such as 'SEC-listed: UBS' or 'not SEC-listed'; set include_unmatched false for listed companies only). The data covers corporate PACs only: trade associations, membership groups (e.g. the American Bankers Association, AICPA, Realtors) and labor unions are not included, refunds are left out, and when you give no company, member, committee, cycle or dates the list covers the last 90 days (with group_by or committee: the current two-year cycle). A gift to a campaign is not a vote and says nothing about intent; FEC reports arrive monthly or quarterly, so the newest weeks are incomplete. Free for every account.
get_pac_donations
One-call briefing on crypto from public filings: trades in cryptocurrency (and in crypto stocks and ETFs) by members of Congress, who holds the spot bitcoin and ether ETFs among 13F filers with the quarter-over-quarter change per fund, which public companies report holding crypto (fair value, cost, year-over-year change) and the weekly CFTC positioning of leveraged funds, asset managers and dealers in CME bitcoin and ether futures, in coins. Each section has source links; a section that fails reports an error without sinking the rest. For a Basic account the answer is the congress crypto trades section; the ETF holders, company treasuries and CME futures sections are replaced by an upgrade message (not_included). The congressional part is free for every account; the other datasets need the Premium plan and a Basic account gets an upgrade message in their place. Rows carry their official source link.
get_crypto_overview
Short sale volume, off-exchange volume and DPI for each stock on one day, from FINRA's daily files. The files cover only trades reported to FINRA instead of to a stock exchange: dark pools and other alternative trading systems, and trades a broker fills itself, such as retail orders filled by a wholesale market maker. off_exchange_total is that volume, not the stock's total trading volume, and the files are not per-venue dark pool volume (FINRA publishes that weekly, with a delay). short_volume is the part sold short; dpi is short_volume divided by off_exchange_total (a ratio, for example 0.46; null when off_exchange_total is 0); dpi_5d and dpi_20d are the volume-weighted DPI over the stock's last 5 and 20 trading days, the day itself included, and are null when the stock was not reported on each of those days. This is one day's short sales, not short interest (positions still open). With no ticker, the stocks with the highest dpi on the latest trading day (or the date you give); listed companies with at least a million shares reported unless you change min_volume or set show to all (which includes funds). With a ticker, its last 30 trading days. A dpi of 0.40 to 0.60 is ordinary for many stocks, because market makers often sell shares they do not yet own to fill customer buy orders; a high dpi does not by itself mean investors are betting against a stock. source_url is FINRA's file for the day. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_short_volume
Estimated net worth of a member of the US House, from the annual financial disclosure (Ethics in Government Act): assets minus liabilities (Schedules A and D), as a LOW, MIDDLE and HIGH figure in US dollars for each report year, with the number of assets and liabilities listed and a link to the report (source_url). The form reports value brackets, not exact figures, so the range is wide by construction: low counts every asset at the bottom of its bracket and every debt at the top, high the reverse, middle the midpoints. With member (Bioguide id or name): every report year we have for them, newest first; with no member: the ranking of sitting House members at their latest report year, highest middle estimate first (rows sets how many, include_former adds former members, year limits to one report year). Estimated from the member's annual disclosure: assets minus liabilities, using the reported value ranges. Excludes the primary residence. Not audited. A row with includes_open_ended_bracket true has an asset or debt in an open-ended bracket ("Over $50,000,000") counted at its lower bound, so the true figure is higher ('at least'). A row with read_from_scanned_report_by_model true comes from a scanned paper report read by a model and can contain errors. A latest row with newer_report_not_read true means the member filed a report for a later year (newer_report_year) that is a scanned paper filing not read yet or could not be parsed, so the estimate is based on the earlier report_year (based_on_note says it). Paid-off debts and the primary residence are left out; when an amendment lists under half as many lines as the full report of its year, the full report is used (report_choice). House members only; Senate reports are not collected. A member with no row has no parsed report yet (the newest reports arrive from May of the year after). This is information, not a valuation or advice. Each row carries its official link (source_url). Free for every account.
get_politician_net_worth
What public companies pay their CEO and other top executives, from the pay-versus-performance table of each proxy statement (DEF 14A), one row per company and fiscal year: the CEO's total pay (Summary Compensation Table), the 'compensation actually paid' under the SEC's formula (it follows the share price and can be far from total pay, or negative), the average of the other named executives on both measures, and the company's and its peer group's total shareholder return (the value of $100 invested at the start of the five-year period). All amounts are US dollars. With no company or dates, each company's latest fiscal year, highest CEO pay first; with a ticker or cik, every year, newest first. Only companies that report a single CEO for the year have CEO figures. amount_to_check is true when a pay amount is over $1 billion, usually a filer tagging thousands as dollars; filing_url is the filing to compare with. Pay says nothing by itself about performance. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_executive_pay
Decisions the FDA has made, from openFDA: approved drug submissions (Drugs@FDA), 510(k) device clearances and PMA device approvals, newest decision first. action_kind is original (the first approval of a drug or device application), supplement (a later approved change, often minor) or clearance (510(k)). With no company and no filters, the first approvals of brand-name drugs and biologics (NDA and BLA) from the last 180 days; new_drugs false, source, application_type or kind select other lists, and a ticker or cik selects everything the FDA decided for that company. The applicant is the company named in the FDA's record; ticker and cik are set only for a confident name match (applicant_match_method says how), and most applicants (generic makers, private device makers) do not match, so a null ticker does not mean the applicant is private. This is a record of decisions already made: it is not a calendar of coming decisions, no review dates (such as PDUFA goal dates) are published, and a drug under review does not appear. The FDA's files can lag a decision by days to weeks. The data comes from openFDA; openFDA asks users not to imply that the FDA endorses their product or service, and this service is not endorsed by or affiliated with the FDA. Not for medical decisions. Information, not investment advice. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_fda_actions
Loans the Federal Housing Administration insures, from HUD's own files, as totals. dataset=forward (home purchase and refinance) and dataset=reverse (reverse mortgages, HECM) give, for a company (ticker or cik) or a lender (the start of its name as HUD prints it), the loans, dollar volume, average rate and share of all loans per month. dataset=multifamily gives the apartment, senior-housing and care-home loans a company holds or services (active projects, balance, recent endorsements, servicing portfolio), or the largest holders when you name none. A joint venture is its own row (joint_venture true) and is never added to the company that part-owns it: it is flagged. The files list no borrower, so there is nothing about individuals. Every answer carries source_url (HUD's page) and a cite string. Until HUD's data is loaded the tool says it is not available yet. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_fha_lending
Short interest from FINRA, published twice a month (settlement dates around the 15th and month end, about a week late): the shares investors have sold short and not yet bought back. With no company, the most shorted stocks in the newest report, ranked by shares short, days to cover (shares short divided by average daily volume; stocks under a million shares short are left out) or change from the previous report. With a ticker or cik, that company's recent reports, newest first. Each row has the current and previous shares short, the percent change as FINRA publishes it, average daily volume, days to cover and a link to FINRA's file. This is open positions, not one day's short selling (daily short volume is a separate dataset). A high figure can reflect bets on a fall or hedges against other positions; it is not a forecast. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_short_interest
Fails to deliver from the SEC's twice-monthly files: for each settlement date, the balance of shares that brokers had failed to deliver to the buyer and not yet delivered, with the previous day's closing price. THESE ARE CUMULATIVE BALANCES, not new fails that day: a share that stays undelivered for ten days is counted on each of the ten days, and the file does not say how old a fail is. The SEC states the numbers are not evidence of abusive or 'naked' short selling and are not necessarily the result of short selling; fails have many causes (processing delays, market makers filling orders, hard-to-borrow shares) and most clear within days. A large balance is a reason to look closer, not a conclusion. With no ticker, the largest balances on the latest settlement date (or the date you give) ranked by value (shares times the SEC's price, the default), shares, share of average daily volume (FINRA's latest short interest report on or before the date) or share of shares outstanding (the company's latest cover-page count; ranked among the 500 largest balances by value); listed companies only unless show is all, because the biggest balances are often in funds. With a ticker, that stock's balance on each recent settlement date. The latest settlement date is usually two to six weeks old. This is not short interest (the short interest page also shows fails beside it) and not daily short volume. source_url is the SEC page that lists the files. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_fails_to_deliver
Lobbying Disclosure Act filings, newest first: the client company, the lobbying firm, what was paid or spent, the issue codes lobbied on and a link to the filing. Totals (income paid to firms, expenses of self-filers) cover all matching filings. Filter by company, year, issue code and dates. With former_government_staff true it returns instead the lobbyists whose filings say they once worked in a member of Congress' office, on a committee or in the executive branch: one row per lobbyist, client company and office, with the office as written, the job title, the years when stated, the lobbying firm, the latest filing and a link to it. A member is linked by Bioguide id only when exactly one member of Congress fits the name, chamber, first name and years; otherwise the name stays as written. Each row reflects what was disclosed at the time of filing, and a lobbyist listed may no longer be registered. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_lobbying
The full document for a document id: a company briefing (company:<cik>), a member of Congress (politician:<bioguide id>), a 13F fund manager (fund:<cik>), or one dataset of a company (congress-trades:<cik>, insider-trades:<cik>, holders:<cik>). Returns id, title, text (plain readable text with a source link on every row), url and metadata. The congressional part is free for every account (members of Congress, Congress trades, the congressional part of a company briefing); the other documents need the Premium plan and a Basic account gets an upgrade message in their place.
fetch
H-1B, H-1B1 and E-3 labor condition applications (LCAs) that the US Department of Labor decided, newest decision first: employer, job title, SOC occupation, wage offered, prevailing wage, worksite and the number of workers asked for. An LCA is a required step before an H-1B petition. It is not a hire, a petition or a visa, and one application can cover several workers (total_workers): say 'H-1B labor applications', never 'H-1B approvals' or 'visas granted'. status is certified, certified_withdrawn (certified, then withdrawn by the employer), denied or withdrawn. With no company, the last 270 days of decisions; give a ticker or cik for one company (every year we hold), or since and until for a range. Wages are as filed with their unit; wage_annual restates the low end of the offered range per year (hour x 2080, week x 52, bi-weekly x 26, month x 12) and is null when the unit is unknown. The employer is matched to a public company by name, so a null ticker does not mean a private employer, and a similar name can be matched wrongly. Most employers are private companies, universities and staffing firms, which file for work at other companies. The files carry no worker names. The Department publishes about six weeks after each quarter ends, so the newest decisions are about two months old. source_url is the Department of Labor's disclosure data page. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_h1b_applications
Which companies are going public: new and upcoming IPOs (this week, this month or any date range), from SEC registration statements (S-1, S-1/A, F-1, F-1/A), final prospectuses (424B4) and withdrawals (RW), one row per registration (file number 333-...), latest activity first. status is filed, amended, priced or withdrawn; is_ipo separates real initial public offerings from resale registrations and follow-ons by listed companies (set ipo true for IPOs only); is_spac marks blank-check companies (spac false hides them). Rows carry the proposed ticker and exchange when the prospectus states them, the maximum registered (offering_amount_usd is the maximum aggregate offering price in the S-1 / F-1 filing-fee table, used to compute the SEC fee: a ceiling, often a round placeholder such as $100 million or $1 billion, NOT the amount raised, and only machine-readable for filings since mid-2025, so it is often null) and, once priced, the amount raised (amount_raised_usd is price_per_share times shares_offered from the 424B4 cover, the base deal without any over-allotment option; null before pricing; price_per_share and shares_offered come from the 424B4 cover), and links to the SEC filing pages. summary, when present, is a short AI-written summary of the latest S-1 / F-1 amendment (bullets on what the company does, headline financials, use of proceeds, who controls it and specific risks, the lead underwriters, and read_from, the filing it was read from with its SEC link); it is information read from the filing, not investment advice, and is a fixed one-line note for a blank-check company. A ticker or cik filter also finds a pre-IPO company by the ticker it proposes. Filters: status, spac, ipo, company and dates (since / until apply to the latest filing). Returns total counts of matching registrations. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_ipo_filings
Transactions by company insiders (officers, directors, 10% owners) reported on SEC Form 4/5, newest first, with insider name and role, shares, price, total value and a link to the SEC filing. By default only open-market purchases and sales (codes P and S) are returned, the usual signal; set open_market to false to include grants, option exercises, tax withholding, gifts. Includes buy/sell counts and dollar totals for all matching rows. A list needs a ticker or cik. group_by 'insider' (or 'company') ranks sellers or buyers in one query, with no paging through rows: it returns the biggest insiders by total dollars with transaction counts and a Form 4 link, over all companies or for one (add ticker or cik), for a date range (since / until; default the last 90 days) and side 'sell' (default) or 'buy'. Example: who sold the most stock this quarter = group_by 'insider', side 'sell', since the first day of the quarter. Totals are shares times price as filed, open-market lines only unless you set txn_code; amendments are left out of rankings. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_insider_trades
Who holds a company among institutional managers filing Form 13F: number of holders, total shares and value, how many managers increased, decreased, opened or closed positions versus the prior quarter, and the largest holders with their change. Or, with manager_cik, the largest positions of one manager. 13F data is quarterly and filed up to 45 days after quarter end; the latest quarter is chosen automatically unless it is still mostly missing. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_institutional_holders
Mass layoff and plant closing notices that employers file with state governments under the WARN Act, newest notice first, from twelve states (CA, KY, NE, NJ, OK, OR, PA, SC, TN, UT, WA, WI). Notices for one employer, state, place, date and type are added up: employees is the workers named and sites the number of lines. Only about one notice in ten is matched to a public company (ticker, cik); with no company given, only matched notices from the last 180 days are listed unless you set listed to false or give dates. New Jersey and Pennsylvania publish no notice date, so those rows only appear when you give a company. employees is null when the state gave no count. A notice is advance warning: the layoff can end up smaller, later or called off. Employer names are matched to companies automatically, so a match can be wrong. source_url is the state's WARN page (the notices are rows on it). Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_layoff_notices
Profile of a member of Congress (party, state, chamber, term), their committee assignments with leadership titles, their most recent reported stock trades and trade totals (count, purchases, sales, most-traded tickers), and a small campaign_finance block for the latest two election cycles (raised, spent, cash on hand, debts, with the date of the last FEC report and a link; the full figures are a separate dataset). Takes a Bioguide id, or a name to look up. Senate trades come from an independent public mirror of the Senate's electronic reports (since 2014, no paper filings); each trade links to the official report. Each row carries its official link (source_url). Free for every account.
get_politician_overview
Early-termination notices the FTC publishes under the Hart-Scott-Rodino Act, newest first: a notice means the parties to a reportable merger or acquisition asked to end the 30-day waiting period early and the FTC and the Department of Justice granted it. COVERAGE IS PARTIAL BY DESIGN. Only granted requests are listed: deals that waited out the full period, deals refused early termination and deals that were never reportable do not appear, so this is not a list of all mergers. The FTC granted no early terminations from 2021-02-04 until the new HSR form took effect on 2025-02-10, so there are no notices dated in between. Each row has the buyer (acquiring_party, the ultimate parent entity), the target (acquired_party) and the entities acquired, and the date granted. Parties are matched to SEC companies by an automatic name match: ticker and cik are set when it matched, and *_match says 'match' (a close name) or 'possible' (a looser one that can be wrong); most parties are private companies, funds or people and have no match. With no company or dates, the last 365 days. A notice is not an endorsement and does not say whether the deal closed. source_url is the FTC page that lists the notices (transaction_number identifies the notice there). Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_merger_clearances
Mergers, tender offers and going-private transactions grouped from SEC filings since 2015, one row per deal, latest filing first. The filings are tender offers (SC TO-T, SC 14D9), going-private statements (SC 13E3), merger registrations (S-4, F-4), merger proxies (PREM14A, DEFM14A, PREM14C, DEFM14C) and deal communications (425). deal_type is tender_offer, going_private, merger or registration (S-4 / F-4 only, which also cover debt exchanges and employee plans); company buybacks (issuer_tender: a company's own tender offer for its shares, mostly funds repurchasing shares every quarter) are left out unless include_buybacks is true or type names them. stage is only the furthest form seen (announced, registered, tender_offer, going_private, preliminary_proxy, final_proxy). subject is the company named as the subject most often, usually the target but not verified (a 425 can be filed by either side; S-4 and F-4 name only the registrant); filer is the other company that files most. Grouping is by rule, not by reading the documents: same SEC file number (005- or 333-), else the same two companies within 270 days, else one company's only open deal. 'grouping' says which was the strongest link; a filing that fits two open deals is left on its own and counted in ambiguous_filings, so one deal can appear as two rows and two deals of the same companies close together can appear as one. What this is NOT: it has no offer price, premium or terms (the filings' headers carry none and the text is not read), and no outcome: the filings never say whether a deal closed or was dropped, and last_filed_date is the latest filing, not a completion date. Information, not advice. A ticker or cik filter finds deals in which the company is named in any filing as subject, filer or co-registrant. Each row has the latest filing's SEC link and the most recent filings of the deal. Returns total counts of matching deals. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_deals
The fund leaderboards from the latest quarter's 13F reports (a 13F lists a manager's long US-listed stock positions, filed up to 45 days after quarter end), and the list of about 75 notable funds. Pick a board: notable-funds (each notable fund with its value, positions, top-10 share, growth, estimated return, turnover and biggest holding); stocks held by the notable funds: most-owned, big-bets, top-buys, top-buys-2q, top-sells, top-sells-2q, new-positions; positions of the notable funds: top-positions, near-52-week-low; the largest single-stock moves of any manager: biggest-increases, biggest-decreases; managers: largest-funds, fastest-growing, concentrated, estimated-performance (set notable_only to limit these, and biggest-increases and biggest-decreases, to the notable funds; without it the moves boards are the 100 largest of all managers). Each row has a source_url: a manager's 13F filings on EDGAR, or for a stock that many managers hold, its 13F.ai page that lists each filing. Estimates, not fund performance. est_return_1q and est_return_4q assume the fund held exactly what it reported at the start of each quarter until the end of it, priced on both dates; a 13F lists long US-listed positions only, so cash, shorts, options, fees and trades inside the quarter are missing. est_copy_return is what buying the earlier report's stocks the day that filing became public and holding to the day the next one became public would have earned. Read all three loosely when turnover_1q is 0.3 or more (the fund trades often) or est_coverage is low (the share of the start-of-quarter value we could price). Returns, shares and ratios are fractions: 0.05 is 5%. One stock's holders and one manager's positions are a separate 13F holders dataset. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_fund_leaderboards
Returns one published theme (a list of companies with something in common, chosen under a stated rule) with its membership rule, review date, change log, the case commonly made for it and the risks commonly cited, comparable ETFs and related themes. A theme is information, not a recommendation: membership is not a forecast and nothing here is advice. For a curated theme each company shows its tier and role and, over the last 12 months, distinct insiders buying and selling on the open market (Form 4 codes P and S), congressional purchases, sales and distinct members, new federal award dollars and lobbying dollars, plus totals across the whole theme; counts are of people and trades, not dollars of insider trading. The rule themes congress-most-bought, insider-buying-clusters and federal-contract-leaders return the current ranked list instead, and a fields note explains each member's value and detail. The rule theme congress-by-party returns party_lists instead of members: four lists (Democrat and Republican, bought and sold), each with the stocks traded by the most different members of that party, the number of trades and the disclosed dollar ranges added up, plus traders, the number of members of each party who traded. An unknown slug returns not_found. For a Basic account: congress-most-bought and congress-by-party are returned in full; a curated theme is returned with its description, rule and companies and only the congressional counts (purchases, sales, distinct members), with the insider, federal-award and lobbying figures replaced by an upgrade message (not_included); insider-buying-clusters and federal-contract-leaders answer with the upgrade message. The congressional part is free for every account; the other datasets need the Premium plan and a Basic account gets an upgrade message in their place. Rows carry their official source link.
get_theme
SEC Form 144 notices, newest first: the notice an insider or other affiliate files before selling restricted or control shares, with the person, their relationship to the company, the shares and the value the filer typed, the broker's expected sale date, whether a Rule 10b5-1 trading plan is named (has_10b5_1_plan) and the sales they report for the last three months. A notice is a plan, not a sale: the person may sell fewer shares, later or not at all, and aggregate_market_value is the filer's own figure, not a trade price. This is information, not investment advice. Each row is linked to the first Form 4 in our data that reports an open-market sale (code S) of the same company by the same person within 90 days (form4_* fields; days_to_form4 is the days from the notice to that filing, days_to_sale to the first sale it reports). sale_reported_in_form4 false means we found none: the person may not have sold, may not have to file a Form 4 (an ordinary affiliate does not), or the Form 4 may not match by CIK or name. form4_matched_by 'name' can join two people with the same name. With no company given, the last 14 days (the feed page's window) unless you give dates. Filters: company, filer_cik, plan, followed (only notices with, or without, a Form 4 sale), min_value, dates (the filing date). Returns total counts of matching notices. Source links: filing_url (the notice) and form4_url. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_planned_insider_sales
Private placements from SEC Form D notices since 2015, one row per Form D or Form D/A, latest first: the issuer (name, state, entity type, industry group, ticker when it is a listed company), every issuer CIK of the offering, the exemption (06b = Rule 506(b), 06c = Rule 506(c)), the security types, the date of first sale, the total offering amount, the amount sold so far, the amount remaining, the number of investors, sales commissions and the executive officers, directors and promoters named, as filed. Amounts are what the issuer typed: unaudited and not checked by the SEC; the issuer can write Indefinite (offering_indefinite true, amount null). An amendment's total_sold and investors_count replace those of the filing before it (same file_number), so the amount raised so far is the latest filing's figure, never a sum over filings. Pooled investment funds (hedge, private equity, venture funds) are about two thirds of the filings: unless pooled_funds is passed or a company is named, this tool leaves them out, as the Private placements page does; pooled_funds true returns only funds. listed true keeps only issuers with an active ticker (needs since, ticker or cik). What this is NOT: a Form D is a notice, not an approval, and does not say whether the money was raised on the stated terms or who the investors are (only how many); related persons are names only, with no holdings. Information, not advice. Every row has filing_url (the filing's index page on sec.gov / EDGAR) and primary_doc_url (the Form D document itself): both are official SEC pages. Returns total counts of matching filings. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_private_placements
Lists the published themes: curated company lists such as AI chips, nuclear and uranium, defense and government stakeholders, plus lists built from our own congressional, insider and federal-contract data, each with slug, kind (curated or rule), group, one-line summary, company count and page URL. A theme is a published list of companies with something in common, chosen under a stated membership rule; it is information, not a recommendation or investment advice. Each theme's members and what the public record shows about them over the last 12 months are available by its slug. Free for every account. The list itself is a catalogue; what a Basic account sees of each theme's data varies by theme.
list_themes
Recorded (roll-call) votes of the House and the Senate, newest first, from 2015: the chamber, Congress, session and roll number, the date, the question put to the chamber (e.g. 'On Passage'), the result, the yea / nay / present / not-voting counts, a description, the measure it was on (bill_label, e.g. 'H.R. 1234', with its title and a congress.gov link when we have the bill) and source_url, the official page of the vote (senate.gov roll call vote page or clerk.house.gov). Rows are ordered by vote_date (the day the chamber voted), newest first, then by roll number, so the first row is the most recent vote. Returns a page of rows plus the number of matching votes. Filter by member (Bioguide id or name: then every row also has that member's position, Yea, Nay, Present or Not Voting, and the reply counts them), by bill (e.g. 'H.R. 1234' or 'S.J.Res. 7', with congress for the Congress of the bill), by position (needs member), chamber, session, words in the question or description (keyword) and dates. Without a bill filter, congress is the Congress of the vote. Only votes on the record are included: voice votes and unanimous consent are not roll calls, so a bill can pass without a row here. Quorum calls, nominations and the election of the Speaker have no bill. A vote on a rule is tied to the rule's resolution, not the bill it makes in order. Senate votes are read from Voteview (UCLA, voteview.com) and can trail the vote by a few days; they have no time of day and leave out the Vice President's tie-breaking vote. roll_number is the chamber's own vote number and restarts every session; voteview_url (Senate rows) is a secondary link that numbers votes across the whole Congress, so its number differs. A vote is not the member's whole position on an issue. Each row carries its official links. Free for every account.
get_votes
The SEC's own enforcement lists, newest first: litigation releases (the SEC's account of a court action: an allegation unless the release itself states a judgment), administrative proceedings (orders of the Commission, many of them settled) and trading suspensions (a pause in trading in one stock, usually ten business days; a pause, not a finding). The lists give the names of the respondents and a release number, not a summary, and we do not store the text of releases: title is those names, summary is a one-line factual description built only from the stored list entry (type, number, date, names, file number, linked documents, suspension period) and says nothing about allegations or outcomes, and url is the release on sec.gov, which says what the SEC alleges or orders. sec.gov often refuses automated readers; url is the release itself. Being named is not a finding about a company, and this is information, not investment advice. Names are matched to companies automatically (companies lists name, cik, ticker and tier for each). match_tier 'confident' is exactly a company's name, legal form included, and never a person's name, a bare one-word name or half of an "A and B" unless that half is a company by its legal form (for a trading suspension, the CIK printed in the order); 'possible' is a similar name that may be a different company: a possible match is not evidence that the company was charged or sued; include_possible adds those matches. Most respondents are people or private firms and match no company. A trading suspension carries the ticker symbol and the suspension period read from the order. With no company given, the last 60 days (the feed page's window) unless you give dates. Filters: company, release_type, include_possible, dates (the release date). Returns total counts of matching releases. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_sec_enforcement
Search 13F.ai for a company (name or ticker), a member of Congress or a 13F fund manager, and for a company also its Congress trades, insider trades and institutional holders. Takes a name or a whole question: the names in it are looked up. Returns up to 10 results, best first, each with id, title, url and a short text; the id identifies a document that can be retrieved in full. Matches names: filters (a date range, an amount, one chamber, a ranking) are not applied. Free for every account.
search
Find a company (by name or ticker), a member of Congress (by name) or an institutional fund manager (by name) and get the identifiers the datasets use: ticker and CIK for companies, Bioguide id for politicians, manager CIK for 13F funds. Forgiving: misspellings ('Pelosy', 'Tubberville'), nicknames ('Bernie Sanders', 'Jon Tester' or 'Jonathan Tester', 'AOC'), brand names ('SpaceX', 'Google', 'Facebook'), former company names and punctuation or suffix noise (Inc, Corp, &, .com) all resolve, so the name needs no correcting first. Free for every account.
search_entities
Forward and reverse stock splits by the day they took effect, newest first. Announced splits that have not taken effect yet are not included. The ratio reads <new shares>-for-<old shares>: 4-for-1 gives four shares for each one held, 1-for-10 is a reverse split that gives one share for ten. Small stock dividends paid in shares are not splits and are left out. With no ticker, the splits of the last 30 days; with a ticker, the stock's whole split history. The list covers every US-traded symbol, including ETFs and foreign companies' over-the-counter shares; cik and company_name are null for symbols we do not list. The splits are worked out from Tiingo daily price data (Market Data from Tiingo.com) and have no public page, so source_url is the 13F.ai page for the split. A split changes the share count and price, not the value of a holding. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_stock_splits
Stock trades disclosed by members of Congress under the STOCK Act, newest transaction first, with the member, party, state, company (ticker, CIK), transaction type, amount range (amount_min to amount_max in USD; amounts are brackets, not exact) and a link to the filing. Returns a page of rows plus total counts of matching trades. One person's trades: member (a name like 'Nancy Pelosi' or a Bioguide id); ticker is only for stock symbols such as NVDA and never takes a name. Filter by member, company (ticker or cik), chamber, transaction type and dates. Disclosures arrive up to ~45 days after the trade. chamber is house or senate. Senate rows come from an independent public mirror of the Senate's electronic reports (the Senate's site blocks automated access): reports filed since 2014, no paper filings, a small share of lines missing, and amendments are not marked; source_url is always the official report. Each row carries its official link (source_url). Free for every account.
get_congress_trades
Stock trades reported by the President, the Vice President, Cabinet members, other Senate-confirmed officials and White House staff on OGE Form 278-T (the White House publishes the President's), newest transaction first: the official (name, position, agency), the asset as the filer printed it, the transaction type, the transaction date, the amount range, the date the report was filed and a link to the official report. Returns a page of rows plus the number of matching trades. Form 278-T is the periodic transaction report that senior executive-branch officials file with the Office of Government Ethics. They must report each trade over $1,000 within 30 days of being notified of it, and no later than 45 days after the trade. Amounts are ranges, not exact figures. Filter by official (words in the name, e.g. 'wright'), company (ticker or cik, for assets matched to a listed company), kind of filer, transaction type and dates. The form has no ticker column: asset_description is the asset as printed and many assets are bonds, funds or private holdings with no ticker (cik is then absent). Amounts are brackets (amount_min to amount_max in USD; amount_max is null for an open-ended bracket), not exact figures. When amount_text is null the form left the amount blank and endnote usually says it was not more than $1,000. transaction_type 'unknown' means the filer left the type blank (not stated). A row with read_from_scan_by_model true comes from a scanned report (the President's) read by a model and can contain errors. Spouse and dependent-child trades are not told apart. Reports can be filed up to 45 days after a trade, so recent trades may be missing. Public records under 5 U.S.C. 13107(c): used here for dissemination to the public, with no advertising. Each row carries its official link (source_url). Free for every account.
get_exec_trades
Federal contract awards from USAspending, most recently modified first: awardee (matched to an SEC company where possible, including via parent company), agency, amount obligated to date (USD), description and a link to the award. Totals cover all matching awards. Filter by company, agency and dates. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_government_contracts
What one institutional manager's 13F report changed against its previous quarter: reported value and position counts, how many positions are new, added to, trimmed and exited, the 10 largest of each by estimated dollars (shares changed times the later price per share), puts and calls kept apart, and a two-to-four sentence summary written by an AI from those changes alone and checked against them (ai_summary; null when none exists). Defaults to the manager's newest report; period (a quarter end, YYYY-MM-DD) selects another. Common shares only; a report with no previous quarter on record, or one that lists far fewer positions than its cover page declares, is returned without a comparison and says why. manager_cik is the manager's SEC CIK. 13F reports are quarterly, filed up to 45 days after quarter end; changes in shares are not always decisions (index tracking, splits and mergers move positions too). Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_fund_filing_changes
A single feed of newly added data across datasets: congressional trades, open-market insider buys and sells, 13D/13G stake filings, 8-K items, lobbying filings, federal contract awards, patent grants, status changes of clinical trials sponsored by public companies and FDA first approvals and device decisions for them. Each event has a one-line summary, the company or member, the event date, when we first stored it and a source link. Filter by companies (tickers), members of Congress (bioguide_ids) and datasets; since defaults to the last 7 days (max 90). For a Basic account only the congress_trades dataset is returned (a request for other datasets alone answers with the upgrade message; a mixed request returns the congressional events and not_included). The congressional part is free for every account; the other datasets need the Premium plan and a Basic account gets an upgrade message in their place. Rows carry their official source link.
get_recent_events
How often the English Wikipedia article about each US-listed company is read, from Wikimedia's page-view data (views from people, all platforms; a day's views arrive about a day late). With no ticker, the company articles read most above their usual level: views over the last 7 days against the article's own daily average over the 90 days before (change_7d of 2.0 means twice as often as usual), or the latest day (change_1d), or the most views (views_7d). Articles averaging under 50 views a day are left out unless you lower min_views. With a ticker, that company's article. Views measure attention for any reason (results, a lawsuit, a product launch, a viral story, a school assignment); they do not say whether the news is good or bad, and some articles cover a brand or a parent group rather than the listed company alone. source_url is the article. Needs the Premium plan; a Basic account gets an upgrade message instead of data.
get_wikipedia_views
13F.ai ChatGPT Plugin FAQ
How the directory, categories and Discoverability Score work.
Read the methodologyHow do I improve 13F.ai's ChatGPT Plugin discoverability?
The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.
Where is this profile measured?
This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.


