Pace

Call this FIRST whenever the user wants to draft, write, or reply to an email to a prospect or customer about a deal. Returns Pace's email-drafting playbook: which CRM tools to call to gather the deal context, the account executive's identity and sending address, and Pace's email-writing guidelines for matching the account executive's style. It does not write the email itself — follow the returned instructions to write the draft.

get_sales_email_playbook

List accounts with optional filters and sorting. Results are paginated (default 50 per page, max 50). Use the `page` parameter to fetch additional pages when `total_matching` exceeds the returned count. Supports filtering by: - team_id: Filter by team/AE group - owner_id: Filter by owner's CRM user ID - custom_filters: Filter by Account custom CRM fields (object_type="Account"). custom_filters format: Each filter is a dict with keys: "field" (CRM API name from discover_custom_fields, e.g. "Region__c" — not the human label), "operator", and "value". Pick an operator compatible with the field's type_category: - string: equals, not_equals, in, not_in, contains, not_contains, starts_with, is_null, is_not_null - picklist: equals, not_equals, in, not_in, is_null, is_not_null - multipicklist: equals, not_equals, in, not_in, contains, not_contains, contains_all, starts_with, is_null, is_not_null - number/currency: equals, not_equals, in, not_in, gt, gte, lt, lte, is_null, is_not_null - date/datetime: equals, not_equals, in, not_in, gt, gte, lt, lte, is_null, is_not_null (value must be ISO-8601) - boolean: equals, not_equals, is_null, is_not_null (value must be true/false) value must be a list for in/not_in/contains_all; omit value (or use null) for is_null/is_not_null; use a single scalar otherwise. Invalid or incompatible entries are dropped before the query runs, so always call discover_custom_fields first to confirm the field name and type_category. Supports sorting by: name, created_at Each account includes an opportunity summary with open_count, closed_won_count, closed_lost_count, total_open_amount, and total_closed_won_amount. Results are scoped based on user permissions.

list_accounts

List opportunities with optional filters and sorting. Use the `status` parameter to filter by opportunity status: - "all" (default): All opportunities regardless of status - "open": Only open opportunities - "closed_won": Only closed-won opportunities - "closed_lost": Only closed-lost opportunities - "closed": All closed opportunities (won and lost) Results are paginated (default 50 per page, max 50). Use the `page` parameter to fetch additional pages when `total_matching` exceeds the returned count. The response includes `page` (current page) and `total_pages` for navigation. Results are always limited to the deals the user is allowed to see. Omit owner_id and team_id to list all of them: the whole organization for users who can see every deal, their own deals otherwise. Supports filtering by: - forecast_category: Rep's forecast (Commit, BestCase, Pipeline, Omitted) - suggested_forecast: AI-suggested forecast - team_id: Filter by team/AE group - owner_id: Filter by owner - forecast_mismatch: True = AI differs from rep forecast - close_date_from/to: Date range (ISO format YYYY-MM-DD) - amount_min/max: Amount range, compared to each deal's amount in its own currency (no conversion) - custom_filters: Filter by Opportunity custom CRM fields (object_type="Opportunity"). - account_custom_filters: Filter by Account custom CRM fields (object_type="Account"). Same dict format as custom_filters. - momentum: AI-assessed deal momentum. Values: positive, neutral, negative - engagement_level: AI-assessed engagement level. Values: high, medium, low, none - forecast_bias: Forecast mismatch direction. Values: over_optimistic, over_pessimistic - win_probability_trend: Trend of win probability over recent evaluations. Values: improving, declining, stable - activity_from / activity_to: Keep only deals whose account had activity in this window (ISO dates YYYY-MM-DD). Both are optional: no activity_from means ever, and activity_to defaults to now and includes the whole day given. "Last 3 days" is activity_from set to today minus 3 days; "this week" is activity_from set to Monday; "never had any activity" is has_activity=false with no dates. - activity_types: Which activity counts. Values: email, call, event (calendar meeting), slack, partner_insight (a partner discussed the deal on a partner call). Omit for all of them. - has_activity: True (default) keeps deals WITH matching activity in the window; False keeps deals with NONE ("open deals with no activity in the last 7 days"). Use these for any "deals with / without activity" question. Never answer one by reading each deal's timeline with get_activities. custom_filters / account_custom_filters format: Each filter is a dict with keys: "field" (CRM API name from discover_custom_fields, e.g. "Region__c" — not the human label), "operator", and "value". Pick an operator compatible with the field's type_category: - string: equals, not_equals, in, not_in, contains, not_contains, starts_with, is_null, is_not_null - picklist: equals, not_equals, in, not_in, is_null, is_not_null - multipicklist: equals, not_equals, in, not_in, contains, not_contains, contains_all, starts_with, is_null, is_not_null - number/currency: equals, not_equals, in, not_in, gt, gte, lt, lte, is_null, is_not_null - date/datetime: equals, not_equals, in, not_in, gt, gte, lt, lte, is_null, is_not_null (value must be ISO-8601) - boolean: equals, not_equals, is_null, is_not_null (value must be true/false) value must be a list for in/not_in/contains_all; omit value (or use null) for is_null/is_not_null; use a single scalar otherwise. Invalid or incompatible entries are dropped before the query runs, so always call discover_custom_fields first to confirm the field name and type_category. Supports sorting by: amount, close_date, name

list_opportunities

Search for accounts by name. Uses fuzzy matching to find accounts even with typos or partial matches. Returns account details including external ID, name, website, domain, and owner. Results are scoped based on user permissions. Users without 'opportunities_see_all' permission will only see accounts they own.

search_accounts

Search organization documents using semantic similarity. Returns the most relevant document chunks matching the query. Useful for finding policies, playbooks, pricing guides, competitive intel, etc. Args: query: Natural language search query describing what to find top_k: Number of relevant chunks to return (default: 5, max: 10) document_id: Optional - filter results to a specific document (by internal ID)

search_documents

Search for emails using semantic similarity. Optionally filter by `opportunity_id` (or `opportunity_external_id`) or `account_external_id` to scope results to a specific deal or company. When neither is provided, searches across all emails in the organization. Unlike keyword search, this finds emails based on meaning, so "budget concerns" would match emails about "cost constraints" or "pricing objections".

search_emails

Search for opportunities by name or account name. Uses fuzzy matching to find opportunities even with typos or partial matches. By default searches across all opportunities (open and closed). Use the `status` parameter to filter results: - "all" (default): Search across all opportunities - "open": Only open opportunities - "closed_won": Only closed-won opportunities - "closed_lost": Only closed-lost opportunities - "closed": All closed opportunities (won and lost) Results are scoped based on user permissions. Users without 'opportunities_see_all' permission will only see their own opportunities.

search_opportunities

Search for teams (AE groups) by name using fuzzy matching. Use this tool to find teams when given a name like "Enterprise" or "West Coast". Also answers "who manages/runs team X?" and shows how teams nest. Returns a list of matching teams, each with: - id, name, and team members - manager: the team's manager ({id, name, email}), or null - parent_team: the team this team nests under ({id, name}), or null If no query is provided (empty string), returns all teams in the organization.

search_teams_by_name

Search for CRM users by name using fuzzy matching. Use this tool to find users when given a name like "John" or "Mary Smith". Handles typos and variations (e.g., "Jon" finds "John"). Also answers "who is X's manager?" and "what does X manage?". Returns a list of matching users, each with: - id, name, email - team: the user's team ({id, name}), or null - manager: the user's direct manager ({id, name}), or null - manages_teams: teams the user manages ([{id, name}]). Only the roots of the managed subtree are listed; managing a root implies managing its sub-teams.

search_users_by_name

Record a new sales forecast entry. Use this tool when a user wants to set the commit amount, best case amount, or add a note to a forecast. Period selection (use `period`): - "default": use the organization's allowed forecast period. Works only if the org forecasts on a single period (quarter OR month). If the org allows both, the tool returns an error and you must specify a period explicitly. - "current_quarter" / "next_quarter": use the current or next fiscal quarter. Rejected if the org doesn't forecast on quarters. - "current_month" / "next_month": use the current or next calendar month. Rejected if the org doesn't forecast on months. Call `get_org_context` if you need to know which periods the org allows. Level selection (inferred from intent — provide at most one of ae_group_id or crm_user_id): - "update my forecast" / "set my numbers" → omit both ae_group_id and crm_user_id (sets the calling user's own personal forecast). - "set [rep name]'s forecast" / "update the commit for [rep] on their behalf" → provide crm_user_id (that rep's personal forecast; resolve the name to an ID with search_users_by_name). - "update the forecast for my team" / "set the team forecast for [team name]" → provide ae_group_id (resolve the team name to an ID with search_teams_by_name). Parameters: - period: One of "default", "current_quarter", "next_quarter", "current_month", "next_month". Defaults to "default", the org's current forecast period. - ae_group_id: Team ID (from search_teams_by_name). Provide for team-level forecasts. - crm_user_id: CRM user ID (from search_users_by_name). Provide to set an individual rep's personal forecast on their behalf; omit to set your own. - commit_amount: The commit amount — the rep's high-confidence number. - best_case_amount: The best case amount — optimistic scenario. - note: Optional note or comment attached to this forecast entry. Permissions: - Your own personal (sales) forecast: requires manage_sales_forecast permission. - Another rep's personal forecast (via crm_user_id): requires manage_reps_forecast permission (sales managers / admins). - Team level: requires manage_team_forecast permission (team managers). Call search_users_by_name / search_teams_by_name first if you need to resolve a name to an ID.

set_forecast

Update one or more CRM fields on a SINGLE opportunity, writing directly to the CRM (Salesforce or HubSpot). This is a WRITE that overwrites existing CRM values. WHEN TO USE: - The user gave an explicit instruction to change a specific field to a specific value on a specific opportunity (e.g. "set Amount to 150000", "change CloseDate to 2025-06-30", "clear Next_Step__c"). WHEN NOT TO USE: - The user only wants to read, list, or analyze data — never write as a side effect of a read. - The field, value, or opportunity is ambiguous — clarify first; do not guess. - You inferred a change yourself — only act on the user's explicit instruction. INPUTS: - Identify the opportunity via opportunity_id (internal ID from search/list tools) OR opportunity_external_id (CRM record ID). - updates: a list of {"field_name": <CRM field API name>, "value": <value>}, e.g.: [ {"field_name": "Amount", "value": 150000.0}, {"field_name": "CloseDate", "value": "2025-06-30"}, {"field_name": "StageName", "value": "Negotiation"}, {"field_name": "Tags__c", "value": ["gold", "priority"]}, {"field_name": "Next_Step__c", "value": null} ] VALUE RULES (provider-agnostic): - Picklist: the option's API value — the KEY in discover_custom_fields "options", NOT the human label. - Multipicklist: a list of API values. - Clear a field: null (or [] for multipicklist) — only allowed when the field is is_nillable. - Number/currency: a number; date: "YYYY-MM-DD". PRECONDITIONS: - Call discover_custom_fields first and confirm the field's "is_updateable" is true (and "is_nillable" is true if you intend to clear it). "is_updateable" reports editability through this tool, which edits only Opportunity fields, so every Account field reports "is_updateable"=false. For picklists, use a value from "options". - Use the exact "name" string from discover_custom_fields for field_name, copied verbatim — do not re-case or guess it. Field API names are provider-specific (e.g. HubSpot uses lowercase "amount", Salesforce uses "Amount"); the examples above are illustrative, not the name for your CRM. BEHAVIOR: - Atomic: if ANY field is ineligible or invalid, nothing is written and an error is returned. - Permissions: you must own the opportunity or have opportunities_edit_all; otherwise it fails. - Returns per-field success/error under "fields_updated", plus any "warnings". Only opportunities are supported. Accounts and other objects cannot be edited via this tool.

update_crm_field