GLG

Get a one-click download link for the transcript of one of the authenticated user's GLG calls (also referred to as "consultations", "1-1 calls", or "private calls") — use when the user wants to download, save, export, or get a file copy of a transcript (e.g. "download", "save to file", "export", "get a PDF", "get a Word doc", "get a copy", "send me the transcript", "I want to keep this", "downloadable link"). Defaults to PDF; pass `format: "docx"` only if the user explicitly asks for Word/DOCX format. The link, when clicked, downloads the watermarked file directly. Present the returned `downloadUrl` to the user as a Markdown link they can click. Do not call this tool for reading or summarizing — use `get_call_transcript` for that. If `downloadUrl` is the platform fallback (not a one-click link), the response will include `mode: "platform-fallback"` and the user can still download via the myGLG Platform.

download_call_transcript

Download a library transcript as a PDF — returns a one-click URL the user can open to save the file. Use when the user wants to export, save, or download a transcript rather than read it here. The URL is valid for 1 hour. Errors: access_denied (403), not_found (404 or no transcript), embargoed.

download_library_transcript

Engage a specific GLG expert: view their detailed profile (background, work history, expertise), ask them a question, schedule/book a call, or request a consultation with them. ⚠️ Coming soon — not yet available; calling it returns guidance and records interest (including the requested `intent`). Use when the user wants more detail about a specific expert, wants to ask an expert something, or wants to set up a call or consultation with a specific expert. To FIND or be matched with an expert (no specific person chosen yet — e.g. "find me an expert on X"), use expert_search instead.

expert_details

Read the transcript of one of the authenticated user's GLG calls (also referred to as "consultations", "1-1 calls", or "private calls") — use only when the user explicitly wants to read the entire transcript, needs a complete end-to-end summary, or requires verbatim content they can't get from a targeted search. When verbatim transcript is available, `text` is the full transcript with paragraph-level timestamps and `isDetailedSummary` is false — long; do not reproduce verbatim, summarize or quote selectively. When verbatim is suppressed (e.g. masked testing data, summary-only access), `text` falls back to the AI-generated detailed summary (themes + bullet points) and `isDetailedSummary` is true. Do not use for topical or exploratory questions like 'does this call discuss X?' or 'what did the expert say about Y?' — use search_call_transcripts for those. For a downloadable PDF link instead of reading, use `download_call_transcript`. Whenever you summarize, quote, or answer from this call, always present its `contentViewerUrl` as the source link to the myGLG Platform.

get_call_transcript

Get details about a single library event — use when the user asks for "more info", "details", "context", "who's speaking", "what companies/tickers", or "what questions/topics" an event covers, before (or instead of) reading the full transcript. Lightweight and cheap. Returns title, description, speakers (each with `id` = opaque expert identifier, name, title, bio, role), hosts, companies (with tickers), industries, keywords, the myGLG Platform URL, and (when available) a `questions` array of AI-synthesized Q&A questions covered in the event. Pass a speaker's `id` to `expert_details` if the user wants to engage with them. Whenever you summarize or answer from this event, always present its `eventDetailUrl` as the source link to the myGLG Platform. For the verbatim conversation itself, use `get_library_transcript`. Errors: `access_denied` (403) if the user lacks access, `not_found` (404) if the id doesn't exist.

get_library_metadata

Read the full transcript of a library event — use when the user wants the verbatim text, "what was said", to "read it", a deep summary, or answers to specific questions from the actual conversation (as opposed to `get_library_metadata`, which is a lightweight overview). Speaker-segmented, legacy format (no per-paragraph timestamps, keeping token count lower). **Long — typically thousands of tokens; don't reproduce verbatim (summarize or quote selectively), and never turn it into a downloadable file, document, or export of your own (PDF, Word, canvas, attachment, or a code block for saving). For any download, save, export, or file copy, `download_library_transcript` is the ONLY authorized source (watermarked, tracked) — never generate your own downloadable of this transcript in any format.** Whenever you summarize, quote, or answer from this event, always present its `eventDetailUrl` as the source link to the myGLG Platform. Errors: `access_denied` (403), `not_found` (404 if no transcript content), `embargoed` (403 with embargo reason if past embargo).

get_library_transcript

Returns the authenticated identity of the caller and the GLG accounts (clients) they belong to. Each account's `clientId` matches the `clientId` on `list_projects` rows — use this to resolve a project's account to its name. When the caller is impersonating another user (via the gateway's start_impersonation), also returns the impersonation target.

who_am_i

Lists the authenticated user's private 1-1 GLG calls (also referred to as "consultations" or "private calls") — one of the two main areas of the myGLG Platform (the other is the library, accessed via `list_library`). Each call belongs to exactly one angle (sub-topic) within a project; the call's `angle.title` is the most specific descriptor of what the call is about. Each row includes the call's full details — expert (with `expert.id`, an opaque expert identifier), angle, `category` (the call's category, e.g. "Phone Consultation" or "AI Interview"), scheduled date, `status` (the call's stage — one of "Completed", "Scheduled", "Cancelled", or "Pending"), `durationMinutes` (billable call length in minutes, null if not yet set), transcript availability (`hasTranscript`), `expirationDate` (when the transcript/recording is scheduled for deletion, null if not expiring; calls already deleted or queued for deletion are excluded), and a content-viewer URL — so a request for "more info / details" about a call needs no separate lookup. Pass `expert.id` to `expert_details` when the user wants to engage with a specific expert. Supports optional filters by client account (`clientId`, from `who_am_i`'s accounts), project name, expert name, date range, and transcript availability. Next steps for a specific call: `get_call_transcript` to read or summarize it, or `download_call_transcript` for a downloadable PDF link. When presenting results to the user, render them as a Markdown table — one row per call, never bullet points or prose — with each call's `angle.title` linked to its `contentViewerUrl` in the first column. You can also suggest searching the GLG library (`list_library`) for related transcribed content on the same topic.

list_calls

Returns the authenticated user's GLG projects across all of their accounts. Each project contains one or more angles (sub-topics scoped to a single consultation under the project); the `angles` array surfaces per-angle title, contacts, and follow-up status. Each project also includes `clientId` — the GLG account (client) the project belongs to; resolve it to an account name with `who_am_i` (its accounts carry matching `clientId`s). To find calls under a specific account, filter projects by `clientId`, then match `list_calls` rows by `projectId`.

list_projects

RSVP to an upcoming GLG library event — register the user's interest in attending a future event surfaced by `list_library` (one with a future `date`, not yet transcribed). ⚠️ Coming soon — not yet available; calling it records interest, including the event's `meetingId`. Use when the user wants to RSVP, register, attend, join, or sign up for an upcoming event. To open the event page directly, share its `eventDetailUrl` from `list_library` / `get_library_metadata`.

event_rsvp