Klue

Search and discover CRM opportunities across the company. Returns one entry per opportunity, deduplicated and ordered by most recent call activity. Each result includes the UP TO DATE CRM state (stage, amount, close date, type), the account name, total call count, and when the last call took place. Use this tool to: - Resolve any account name, account shorthand, deal name, or descriptive deal reference into one or more opportunity IDs before calling the other deal tools (get-buyer-requirements, get-competitors, get-recent-deal-tips, get-deal-tip) - Find deals for a specific account before a meeting - List all open opportunities a seller is working - Explore recently active deals within a time window

deals_find-opportunities

Retrieve what the buyer needs, as expressed across all sales calls for an opportunity. Requires a CRM opportunity ID. If you don't have that, call find-opportunities first to resolve. Each requirement is extracted directly from call transcripts and includes a relevance score (higher = more important to the buyer) and the supporting transcript exchanges that evidence the requirement. Results are ordered by most recent call first, then by score descending. Use this tool to: - Understand the prospect's priorities before a follow-up call or proposal - Identify unmet needs that competitors may be addressing - Tailor messaging to what the buyer has explicitly stated they care about

deals_get-buyer-requirements

Identify which competing vendors or products have been mentioned across all calls for an opportunity. Requires a CRM opportunity ID. If you don't have that, use find-opportunities first to resolve. Returns a deduplicated list of competitors with threat level, and first/most-recently-seen timestamps across the full call history. Deduplication uses org_unit_id when available, falling back to rival_id or product name. Use this tool to: - Understand the competitive landscape for a specific deal - See which competitors have emerged recently vs. throughout the cycle - Prepare competitive positioning before a call or proposal

deals_get-competitors

Fetch the complete content of a single deal tip for one sales call. Requires a deal_tip_id (UUID) obtained from get-recent-deal-tips. Always call get-recent-deal-tips first to retrieve a valid deal_tip_id — never pass an opportunity ID, account name, deal name, or any other non-UUID reference here. Returns the full AI-generated email (subject + body), deal summary, all extracted buyer requirements, detected competitors, and call metadata (title, URL, source, participants). Sellers are automatically restricted to tips from calls they participated in. Use this tool to: - Get the full detail of a specific call after browsing the timeline with get-recent-deal-tips - Read the AI-drafted follow-up email and deal summary for a call - Pull the complete buyer requirements and competitor list for a single call

deals_get-deal-tip

Get a lightweight chronological timeline of AI-generated deal tips for an opportunity. Requires a CRM opportunity ID. If you don't have that, call find-opportunities first to resolve. Returns one summary per sales call, ordered most recent first. Each entry includes the AI deal summary, email subject, CRM context at the time of the call, and call participants. This is the right starting point for exploring a deal's history. Use the deal_tip_id from results to fetch full details — email content, buyer requirements, and competitors — via get-deal-tip. Use this tool to: - Quickly scan what has happened across recent calls on a deal - Find a specific call to drill into with get-deal-tip - Summarise deal progress for a manager or account review

deals_get-recent-deal-tips

Match a seller question to Klue Smart Answers. Returns the most relevant Smart Answers, most relevant first. Returns an empty list if no relevant Smart Answer is found. Parameters: - user_question: The seller's question to match. - top_k: Maximum number of Smart Answers to return (default: 3).

klue-answers_match-smart-answers

Report that competitive data you needed was missing, incomplete, or unreachable. You SHOULD call this tool whenever, while answering a competitive question, any of the following happened: - You looked for data or an artifact type that does not exist in Klue (no results anywhere for the competitor, deal, or topic). - Searches returned incomplete or thin results, so your answer rests on partial coverage. - You had to approximate or answer from adjacent data because the specific thing asked for was missing. - No available tool or filter could express what you needed to look up. Call it AFTER you have finished retrieving (so what_i_tried is complete) and BEFORE you give your final answer. Calling this tool is expected behavior, not a failure — it is how Klue learns which competitive data its users need but do not have. It does not change your answer: still answer the user with the best available data and state what was missing. Do NOT call it when retrieval fully answered the question, or when you have not yet widened the search per the fallback rule — exhaust the obvious retrieval paths first.

klue-content_report_data_gap

Search Klue's competitive intelligence for a competitor, account, or topic. Covers battlecards and cards, agent insights, evidence quotes from buyer conversations, win-loss interviews, and Blindspots buyer/market research interviews. Each result is a single passage with a ``category`` label and a ``public_url`` to cite. Example questions: - "What objections come up around pricing?" - "Find anything tied to the Acme opportunity." - "What did buyers say about competitors this quarter?"

klue-content_search_klue_content

Search win/loss interviews — reports and verbatim transcripts. Reports are the analyst-written narrative summarising why a deal was won or lost; transcripts are the raw interview content. When to use: - Questions about deal outcomes, competitor displacement, pricing pressure, or loss reasons. - Use ``subcategory='report'`` for synthesised insight, ``subcategory='transcript'`` for exact quotes. Other tools: prefer this when the question is about win/loss specifically; ``search_klue_content`` for cross-source recall. For accurate interview counts or an interview summary, use ``alpha_read_search_win_loss_documents_v1``. Each result is a single passage with a ``public_url`` for citation. Example questions: - "Why did we lose the Acme deal?" - "What did buyers cite as the top reason for losing to Snowflake last quarter?" - "Find transcripts where pricing came up in losses to Databricks."

klue-content_alpha_read_search_win_loss_v1

Search win/loss interviews (reports + transcripts) — one row per interview, drawn from each document's **generated summary**, NOT the full text. Use this when you need: - **An interview summary.** The result's content IS the generated summary of the interview — surface it directly when the user asks for a synopsis or "what did this interview say". - **Accurate interview counts.** ``total_count`` reflects the number of matching interviews. ``alpha_read_search_win_loss_v1`` counts chunks and can't give you a real interview count. - A list / overview of relevant interviews before pulling detail (verbatim quotes, full analyses) via ``alpha_read_search_win_loss_v1``. ``query_text`` is optional — omit for a filter-driven listing, or supply it to rank by hybrid relevance against each document's summary. Example questions: - "How many losses to Snowflake did we record last quarter?" - "List every win/loss report for deals over $250k." - "Which interviews involved Databricks as a competitor?"

klue-content_alpha_read_search_win_loss_documents_v1