Alium

Instructions and this week's searches for the user's scheduled Alium weekly report. Call it when running that scheduled task (or when the user asks for their Alium weekly report), then follow what it returns.

weekly_report

List the valid industry and use-case tag values accepted by `search`. Use this before searching when the user mentions an industry or use case in vague terms — confirm the canonical name first, then filter. Example: user says 'retail' → call this with `kind: 'industries'`, find 'Retail > Footwear', then pass that exact value to `search` as the `industry` filter. Returns labels sorted by frequency with usage counts.

list_taxonomy

Find buyer accounts (companies) you have interviews on. Matches the transcript filter surface (industry, in_market_for, vendors, products, ratings, revenue, employees) but deduplicates results to one row per account. Decompose the request into structured filters instead of putting it all in `q`: `in_market_for` for use cases, `industry` for verticals, `vendors`/`products` for tools, `revenue`/`employee_count` for company size. Reserve `q` for free-text concepts that don't map to a filter. Example: "accounts in market for CRM with revenue over $1M" → in_market_for: "CRM", revenue: "1000000;". Returns account slugs you can pass to `fetch_account` for full account-level Q&A, tech stack, and ICP context.

search_accounts

Find buyer interview transcripts. Use this when the user asks about who said what, what buyers in industry X think, who's evaluating product Y, who's dissatisfied with vendor Z, or which companies are in market for a use case. Decompose the request into structured filters instead of putting it all in `q`: use `in_market_for` for use cases, `industry` for verticals, `vendors`/`products` for tools, `buyers` for the companies being interviewed, `revenue`/`employee_count` for company size, `min_rating`/`max_rating` with `products` or `tech_stack_category` for satisfaction. Reserve `q` for free-text concepts that don't map to a filter. Example: "in market for CRM at companies with revenue over $1M" → in_market_for: "CRM", revenue: "1000000;". Returns summaries with transcript IDs you can pass to `fetch` for full details. Each summary already has the buyer's industry, revenue and employee bands, location and intro status, so don't `fetch` every result just to compare or list them.

search

Find vendors (companies) that appear in your team's interviews. Use this when the user asks about a vendor — competitor analysis, customer base, who buys X — and you don't yet have the vendor's slug. Returns vendors ranked by transcript volume with slugs to pass to `fetch_vendor`.

search_vendors

List the intros on the caller's Alium Intros tab: a team admin sees every intro on the team, anyone else the intros they take. Each has its status as the app shows it, the buyer company, who requested and who takes it, the meeting time, whether the recording is ready to download from the intro page, and the linked interview. Filter by `status` (upcoming / past) or by buyer `account` name.

my_intros

Read the full text of an Alium help article by its docs.alium.io URL (from `search_help` results). A URL with a #section anchor returns just that section; without one, the whole article.

fetch_help

Read a buyer account by slug — KPIs (interview count, last interviewed), ICP context (revenue/employee/industry match), tech stack aggregates, and the account-level Quick Study (synthesized answers to standard questions across every interview at the account). Use this after `search_accounts` returns a slug. Pass `fields` to slim the response (the Quick Study is large, so it is omitted by default); pass ['all'] for everything.

fetch_account

Read one interview by transcript ID, after `search` returns an ID or whenever the user asks about a specific interview. Vendor users get the executive summary, priorities, in-market tags, and tech stack; two larger sections are omitted unless requested: 'transcript' (the actual interview, interviewer/interviewee turns, use for exact wording and quotes) and 'quick_study' (synthesized answers to standard questions, not verbatim). Buyer users get the redacted transcript paragraphs and tech stack as JSON. For quotes on a topic, pass grep: for vendor users it searches every section, the verbatim transcript included; for buyer users it searches the interview paragraphs. Either way only the matching lines come back.

fetch

Read a vendor's full profile by slug — overall ratings, top buyers, competitive overlap, what buyers like and dislike. Use this after `search_vendors` returns a slug, or whenever the user wants competitive intelligence on a specific software company. Optional `product` and `use_case` arguments scope the profile to one product line or use case.

fetch_vendor