In Practise

Adds companies, moats, business models, or industries to the user's watchlist. This tool allows adding multiple items at once. Each item requires: - type: One of 'company', 'moat', 'businessModel', 'industry' (optional, defaults to 'company') - idOrName: Either a UUID (for direct ID lookup) or a name (for fuzzy matching) For companies: Accepts company names, display names, or UUIDs For moats: Accepts moat names or UUIDs For business models: Accepts business model names or UUIDs For industries: Accepts industry names or industry codes (numbers) The tool handles both exact UUID matches and name-based lookups. When adding moats, business models, or industries, it will also automatically restore any previously deleted companies that are associated with those categories. Returns a summary of how many items were successfully added in each category.

add_to_watchlist

Compare the SAME section between two filings of one company (e.g. this year vs last) to answer "what changed in their risk factors / accounting policy / MD&A?". Params: filingA (earlier filing UUID), filingB (later filing UUID) — resolve BOTH with forensics_list_filings; sectionType (one of the fixed section names). Returns: { present, sectionType, a/b: {filingId, citation}, granularity ("item" | "block"), added: [{status: "genuinely_new", heading, page, text}], removed: [{status: "genuinely_removed", ...}], changed: [{status: "reworded_carried_over", heading, page, oldText, newText, similarity}], unchanged: count, addedTotal/removedTotal/changedTotal, truncated, note }. When granularity="item" (sections with headings: risk factors, notes, MD&A) each entry is a whole SECTION ITEM named by its `heading`, so the counts are ITEMS (e.g. "2 new risk factors, 5 reworded"), not blocks. "block" is the fallback for heading-less sections. An item is `added` only if ALL its content is new, `removed` if all gone, else `changed` — an item that persists but is reworded (including a restructured item: a heading in one filing, a paragraph in the other) is classified `changed`, not added/removed. Within a `changed` item, oldText vs newText carry the wording delta: a concept carried over from the earlier filing stays in oldText even when the surrounding wording changed (forensics_find_mentions confirms whether a given phrase is truly absent from the earlier filing). When truncated the lists are capped (the totals are reported). Verbatim passages come from added/removed/changed, cited to a and b. Requires forensic-accounting access.

forensics_section_diff

EXHAUSTIVE literal-substring scan over a company's SEC/financial filing blocks. Returns EVERY block whose verbatim text contains `term` (case-insensitive substring) — COMPLETE RECALL, no relevance ranking, no silent cap. USE THIS (not forensics_search_filing_text) for "find ALL / every / list all mentions of X", related-party or counterparty name sweeps (e.g. every place "DriveTime" appears), covenant/entity occurrence audits, and any question that must not miss a differently-phrased mention. forensics_search_filing_text is RANKED and CAPPED — it returns the most relevant passages, NOT all of them, and will silently drop long-tail mentions. First resolve the company with search_companies to get its `companyId` (a UUID). Omit `filingId` to sweep every filing (each match is labelled with its `filingLabel`), or pass one to scope (e.g. the latest 10-K). Return fields: - `totalMatches`: the EXACT number of matching blocks across everything scanned — this is the recall guarantee. - `filingsScanned` / `filingsTotal` / `filingsWithoutText` + `coverageWindow`: the searched set vs the company's full filing record in scope. A result of 0 matches is a valid negative ONLY for the `filingsScanned` filings; `filingsWithoutText` filings have no ingested text and were NOT searched — a mention there is invisible to this tool. When filingsWithoutText is 0, 0 matches bounds the company's entire filing record in scope. - `matches[]`: each = `filingLabel`, `sectionType`, `page`, ltree `path`, `occurrencesInBlock`, a verbatim `snippet`, and a `deepLink` that opens the filing viewer at that exact block (scrolls to it and highlights the passage). In responseFormat "detailed", the first matches also carry the full verbatim block `markdown`. - `note`: if set, the location list was capped (very common term) — narrow by `filingId`/`sectionTypes`. `totalMatches` is still exact. `snippet`/`markdown` are VERBATIM block text; forensics_verify_quote grades whether a specific figure is an exact substring. Matching is against literal stored markdown, so a term split by inline markup (e.g. "Drive**Time**") is not matched. ACRONYM / DEFINED-TERM CAVEAT: this is a LITERAL matcher, so a multi-word term that a filing defines once and then abbreviates is UNDERCOUNTED. A document that writes "artificial intelligence ("AI")" once and then says "AI" thereafter returns ~1 match for "artificial intelligence" while the topic saturates the section. A low `totalMatches` on a spelled-out phrase is therefore NOT evidence the topic is rare. For any term with a common abbreviation (artificial intelligence↔AI, remaining performance obligation↔RPO, material weakness↔MW), run BOTH forms and sum `totalMatches`; the short form usually carries the true count.

forensics_find_mentions

Retrieves specific InPractise articles by their URL slugs. Interviews return complete content. Analysis, Weekly Update, and Company Profile results are passage-scoped: copy both the slug and passageId from search_articles into `slugs` and `passages`; without that passageId the aggregator is withheld under `provenanceRequiredSlugs`. Copy an aggregator article's returned `citation` in full so its source-interview link stays attached. Other access and subscription-window restrictions are returned under `inaccessibleSlugs` and `windowRestricted`; relay any `accessNote` or window message.

get_articles_by_slug

Retrieves essential platform data that provides context for all subsequent queries. This tool should typically be called first to understand the current state of the platform and available data. Returns: - User's complete watchlist (companies, industries, moats, business models) with full metadata - Organization member watchlists (if user belongs to an organization) - All available moats with company and article counts (sorted by article count) - All available business models with company and article counts (sorted by article count) - Companies with coverage grouped into 4 categories (only companies with 11+ articles included): * "50+ articles": Companies with 50 or more articles (ordered by coverage, highest first) * "30-49 articles": Companies with 30-49 articles (ordered by coverage, highest first) * "20-29 articles": Companies with 20-29 articles (ordered by coverage, highest first) * "11-19 articles": Companies with 11-19 articles (ordered by coverage, highest first) Each group contains company symbols (fmp_symbol), or ticker if no symbol, or company name if neither is available. - Latest 30 articles with metadata (title, companies, expert info, publication date, etc.) This comprehensive dataset helps understand what companies, topics, and content are most relevant to both the user specifically and the platform overall. The response also reports `accessTier`. On the free tier it includes an `accessNote` explaining what the connection can and cannot read — ALWAYS relay that note to the user early in the conversation.

get_core_data

Fetches the VERBATIM block(s) at a specific section path of a filing — for navigating to a known disclosure or pulling the exact text behind a search hit / flag. Inputs: `filingId` and `path` (the ltree `sectionPath` that forensics_search_filing_text returns, or a path you want to drill into). Returns the section's blocks in document order with their verbatim `markdown`, plus the concatenated section `markdown`. Transcript blocks (earnings calls, event transcripts) additionally carry `speaker` ("Name — Role, Company", null where the source maps no speakers) and `startTimestamp` (the turn's start offset in seconds from the start of the recording); their `page` is null. Use this to (a) jump straight to a disclosure already located, or (b) read the precise source text for a figure. The `markdown` is byte-for-byte from the filing.

forensics_get_filing_section

Resolves a forensic flag to the VERBATIM filing block(s) it was anchored to — the drill-down from a flag (from forensics_search_flags) to its source disclosure. Inputs: `flagId` (from forensics_search_flags), and optionally `sourceTable` (also from search_flags — lets it skip the table scan). Returns the verbatim `sections` (heading + body blocks) the flag points at, the `filingId`, a `deepLink` that opens the filing viewer at the source passage (scrolls to it and highlights it), and `rawQuoteHint`. For policy_diff flags it returns BOTH the current- and prior-filing blocks (the YoY diff). `rawQuoteHint` is the In Practise generator's LOCATING hint, not authoritative source text; the verbatim `sections.markdown` is the quotable content (forensics_verify_quote grades whether a given string is exact). If `sections` is empty (ratio/extract areas have no block anchor, or a stale path), the `note` field says so — fall back to forensics_search_filing_text.

forensics_get_evidence

List a company's SEC/financial filings — each with its `filingId` (UUID), `filingType`, `fiscalYear`, `fiscalPeriod`, `reportDate`, `filingDate`, a human `label`, `hasBlocks` (whether its text is ingested), `sectionTypes` (the DISTINCT section tags of its ingested blocks), and `attributionStatus` (filing-level speaker-coverage flag: 'limited' = some turns have no resolved speaker — a blank or generic "Speaker N" label; 'unavailable' = no speaker mapping at all; null = fully mapped or not a transcript. It is a COVERAGE signal, not per-quote confidence — a quote whose block has a named speaker is attributed even when the filing is 'limited'. Operator/host turns don't count as degraded). Newest first. `reportDate` is the period-end date; `filingDate` is the publication/event date — some sources (e.g. earnings-call and event transcripts) carry only `filingDate`. `sectionTypes` makes the list an event index as well as a filing index: an 8-K carrying `non_reliance` is a restatement notice, `auditor_change` an Item 4.01, `listing_deficiency` an Item 3.01, `exit_costs` a restructuring, `bankruptcy` an Item 1.03; transcripts carry `prepared_remarks`/`qa`/`discussion`. Because every ingested filing lists its tags, the enumeration is also a checked denominator — a company whose listed current reports include no `non_reliance` tag has zero ingested non-reliance disclosures, which is a stronger statement than a search returning nothing. This denominator counts ONLY `hasBlocks: true` filings: a filing with `hasBlocks: false` has UNKNOWN section content (not yet ingested — its `sectionTypes` is empty because nothing was parsed, NOT because it is event-free), so it must be excluded from any "no such event" conclusion. `coverage.filingsWithoutBlocks` counts them; an event screen that treats an un-ingested filing as clean produces a false all-clear. Accepts one `companyId`, or `companyIds` — up to 50 companies enumerated in a single call (e.g. a whole portfolio), returning one result object per company with its `ticker`/`name` and the same filing fields. Also enumerates ACROSS companies when `dateFrom`/`dateTo` are supplied without any company: the date window is what bounds the query, so "which companies filed an earnings release yesterday" or "every restatement notice this quarter" is one call rather than one call per known ticker. `dateFrom`/`dateTo` match the PUBLICATION date (when the filing became public), not the period it covers — a FY2025 10-K filed 2026-02-01 falls in 2026. In this mode the response carries `scope: "all_companies"` and each filing carries its own `companyId`/`ticker`/`companyName`; there is no `coverage` block, which is a per-company statement. A call with neither a company nor a date window is an error, not a full-corpus dump. `sectionTypes` filters to filings carrying at least one ingested block with those tags — the same vocabulary the response reports. It is how the section plane becomes queryable rather than only readable: `sectionTypes: ['earnings_release']` returns the 8-K-exhibit earnings releases described below, `['non_reliance']` the restatement notices. Because tags exist only on ingested blocks, this filter implies `hasBlocks: true`. Resolves a specific filing to its `filingId` — e.g. "the FY2024 10-K", "the latest 10-Q", "the FY2023 annual report" — for use with forensics_section_diff, forensics_get_filing_section, or forensics_find_mentions. This is the reliable filing→UUID resolver; ranked forensics_search_filing_text won't consistently surface a specific period. `companyId` is a UUID as returned by search_companies. `filingType` optionally narrows by exact type; an unrecognised value is an error, never an empty result. When a filingType filter matches nothing but the same-named content exists as SECTIONS of another filing type (US earnings releases are 8-K exhibits: filingType "current_report" with `earnings_release` section blocks), the response says so in `sectionTypeOverlap` — a zero here is checked against the section plane, not just the filing-type column. Each result carries a `coverage` summary: `filingsWithoutBlocks`, `latestFilingDate` vs `latestBlockBackedFilingDate`. When they differ, the newest filing(s) exist but their text is not yet ingested — disclosures newer than `latestBlockBackedFilingDate` may exist that text search cannot see. Filing-type vocabulary: a "10-K" is stored as filingType "annual_report"; a "10-Q" as "quarterly_report"; an "8-K" as "current_report"; a Form 12b-25 (NT) as "late_filing_notification"; "FY2024" corresponds to fiscalYear 2024. Only filings with `hasBlocks: true` have searchable/diffable text. EARNINGS RELEASES are bimodal — do NOT reach them with filingType="earnings_release". A US company's quarterly earnings release is Exhibit 99.1 of an 8-K, ingested as `section_type='earnings_release'` blocks INSIDE a `current_report` filing (thousands of filings); only a minority of (mostly non-US) issuers file a standalone "earnings_release" filing type. So filtering filingType="earnings_release" returns ONLY the standalone minority and misses every 8-K-exhibit release — it will look like "no earnings release in corpus" when the current, ingested release is right there. To find earnings releases: pass `sectionTypes: ['earnings_release']` here (with a `companyId` for one issuer, or with a date window for every issuer that released in that period), or use `forensics_search_filing_text` with the same `sectionTypes` when you want the passage text rather than the filing list. (Earnings presentations behave the same way.)

forensics_list_filings

Removes companies, moats, business models, or industries from the user's watchlist. This tool allows removing multiple items at once. Each item requires: - type: One of 'company', 'moat', 'businessModel', 'industry' (optional, defaults to 'company') - idOrName: Either a UUID (for direct ID lookup) or a name (for fuzzy matching) For companies: Accepts company names, display names, or UUIDs. Companies are marked as deleted rather than fully removed. For moats: Accepts moat names or UUIDs and fully removes the association For business models: Accepts business model names or UUIDs and fully removes the association For industries: Accepts industry names or industry codes (numbers) and fully removes the association The tool handles both exact UUID matches and name-based lookups. For companies, the deletion is soft (marked as deleted) to preserve history, while moats, business models, and industries are hard deleted from the user's watchlist. Returns a summary of how many items were successfully removed in each category.

remove_from_watchlist

Performs semantic search across the entire InPractise article library. This tool searches through thousands of business research articles, interviews with executives, and analysis pieces. It returns the particular matched section of the article content. One common workflow is to query the search with a company name and then follow up queries with different companies, moats, industries, etc, as it makes sense. You can also search by company symbol. Matching is case-insensitive and prioritizes exact tickers, then exact FMP symbols. Only when neither exact form exists does a bare ticker expand to an exchange-qualified FMP symbol such as CSU.TO. You can filter by a company symbol and leave the query as an empty string or undefined to get all articles for that company in descending order of published date. Any requested symbols that cannot be resolved are returned in the unresolvedSymbols field. If none of the requested symbols resolve, the tool returns no articles and never falls back to an unfiltered search. In order to get all the articles for a query, you can get 50 articles and then query again with the next page number. However, you would only want to do this if the user explicitly asks for all the articles, usually for a company. Articles the calling user does not have access to are filtered out of the results and listed under `inaccessibleSlugs`. When the response carries an `accessNote` (free-tier connections), ALWAYS relay it to the user — it explains that results were hidden, why, and how to get full access; `hiddenResultCount` says how many matches were withheld. Quote experts VERBATIM, don't summarise: copy from a result's `content`, render as Markdown blockquotes, and attribute each with that result's `citation` field copied as-is. Each citation deep-links to the exact paragraph the matched content came from (the article page scrolls to and highlights it). To read an interview in full, call get_articles_by_slug with its slug. For an aggregator, copy both its slug and `passageId` into get_articles_by_slug; that keeps the read limited to the verified passage. By default (no `types` specified) results cover expert interviews plus Analysis, Weekly Update and Company Profile passages whose exact source interview passage is currently published and accessible. Aggregator results carry a `citation` containing BOTH the aggregator paragraph and source-interview paragraph deep links, plus structured `provenance`; copy the full citation as-is. Aggregator passages without a valid accessible chain are withheld and listed under `provenanceExcludedSlugs`. Other secondary types remain opt-in through `types`. These are all the available types: - ip_analysis: Research pieces, thoughtful analysis and synthesis of interviews, and deep dives - ip_resource: Research pieces, less deep than ip_analysis - ip_hosted: Interviews conducted by In Practise - 3p_interview: Interviews conducted by IP Partners - ip_dialogue: Dialogues with investors, not used anymore - ip_weekly_update: Weekly update with a summary of the latest published interviews and research - ip_company_profile: Deep dives on a company - ip_value_chain: Company value chain analysis - ip_podcast: Podcasts to educate analyst on the art of primary research. Usually published on the IP Fieldwork podcast on Spotify and Apple Podcasts. - ip_survey: Results of industry surveys conducted by In Practise

search_articles