tenant_ai_call_context_preview
ChatGPTPreview the assembled AI call context (final prompt, first/second message, qualifying questions and pipeline config) for the active tenant, exactly as it would be built for a real call — without placing one. Useful for verifying the effect of intro, company-variable and voice-prompt changes. Set call_type="web" to preview a browser web call instead of a phone call: that forces the outbound pipeline and resolves the intro the way WebCallDispatchService does. Credit score values are masked in previews ("[masked]"); masked_attributes lists which were masked.
tenant_ai_call_history
ChatGPTRead the call history of the active workspace as the Statistics → History screen shows it. summary: total calls, totals by AI result (list_ai_tag) and by failure reason (list_fail_reasons) for a period. list: one page of calls with contact, result and AI summary, no transcript. get: one call with its transcript, summary, questions and answers and a temporary recording link. Pick a period preset; use period=custom with start_date and end_date only when the user names dates. Every answer echoes the window used. get is audited.
tenant_ai_call_history_export
ChatGPTQueue, follow and list spreadsheet exports of the call history of the active workspace, the same file the Statistics → History screen exports, with every column including the transcript. export: queue one for a period and filters and get an export_request_id (one export at a time per user). export_status: progress, and when completed a download link valid for 30 minutes, refreshed on every call. list_exports: the last exports of the workspace. export_status and list_exports cover the exports you requested; with the tenant_users.view permission, every member's. For reading calls or transcripts in the conversation use tenant_ai_call_history instead.
tenant_ai_company_variables
ChatGPTRead the active tenant's AI company variables (the {placeholder} key/value pairs substituted into AI call prompts and messages). Action: read. To change or remove them use tenant_ai_company_variables_update.
tenant_ai_company_variables_update
ChatGPTAdd, change or remove the active tenant's AI company variables (the {placeholder} key/value pairs substituted into AI call prompts and messages). Actions: preview_update, apply_update, delete. apply_update and delete require confirmation="APPLY_TENANT_AI_COMPANY_VARIABLES_UPDATE". Read the current values with tenant_ai_company_variables. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_first_day_followup
ChatGPTRead the tenant first-day followup sequence (the ordered call/message/mms actions with relative delays, and the linked intro). Action: read. Distinct from tenant_ai_followup. To change it use tenant_ai_first_day_followup_update.
tenant_ai_first_day_followup_update
ChatGPTChange the tenant first-day followup sequence. Actions: preview_update, apply_update. apply_update accepts "actions" (the full ordered list of call/message/mms actions) and "ai_intro_id"; supplying "actions" REPLACES the entire sequence. Distinct from tenant_ai_followup_update. apply_update requires confirmation="APPLY_TENANT_AI_FIRST_DAY_FOLLOWUP_UPDATE". Read the current sequence with tenant_ai_first_day_followup. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_followup
ChatGPTRead the tenant followup sequence: the routine (call/message/mms actions) plus the totals and working-hours settings. Action: read. To change it use tenant_ai_followup_update.
tenant_ai_followup_update
ChatGPTChange the tenant followup sequence. Actions: preview_update, apply_update. apply_update accepts "scalars" (totals + working hours) and/or "routine" (the full ordered list of call/message/mms actions); supplying "routine" REPLACES the entire routine. apply_update requires confirmation="APPLY_TENANT_AI_FOLLOWUP_UPDATE". Read the current sequence with tenant_ai_followup. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_intros
ChatGPTList and read the active tenant's AI intros (prompt content, opening sentences, qualifying questions). Actions: list, read (by "id"). To create, change or delete an intro use tenant_ai_intros_update.
tenant_ai_intros_update
ChatGPTCreate, change or delete the active tenant's AI intros. Actions: preview_update, apply_update, delete. apply_update with no "id" creates a new intro; with an "id" it updates that intro. apply_update and delete require confirmation="APPLY_TENANT_AI_INTROS_UPDATE". List and read intros with tenant_ai_intros. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_onboarding_guide
ChatGPTReturns the onboarding guide for the active tenant class (telephony or Talk Kit web calling): the platform gates and confirmation strings, the maintained onboarding checklist from the public prompting guide (in order), the questions to settle with the user before the first change, and the prompt-writing conventions. Useful before configuring a tenant for the first time.
tenant_ai_readiness_check
ChatGPTRun the readiness checks the app runs before a blast starts, without changing anything. Returns {is_valid, errors, warnings}: errors block a blast, warnings do not. Without arguments it checks the tenant configuration (Voice AI phone number, SIP provider, Aegis workflow id) and, for call-transfer tenants, the after-hours intro and working hours. With blast_id it also checks that blast's intro and its contacts' ZIP coverage; intro_id overrides the blast's intro. A blast_id that is not one of this workspace's blasts returns not_found. The Aegis workflow check is a remote lookup whose result is cached.
tenant_ai_settings
ChatGPTRead the allowlisted AI-related settings of the active tenant, with their schema. Action: read. Sensitive values (credentials) come back masked and can only be changed in the BiglySales app. To change other settings use tenant_ai_settings_update.
tenant_ai_settings_update
ChatGPTChange allowlisted AI-related settings of the active tenant. Credential settings (tokens and keys) are read-only here and are changed in the BiglySales app. Actions: preview_update, apply_update. apply_update requires confirmation="APPLY_TENANT_AI_SETTINGS_UPDATE". Read the current values with tenant_ai_settings. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_sinch_numbers
ChatGPTList the active tenant's Telephony Sinch outbound numbers and check which area codes of a US state have numbers to rent. Actions: list, availability (by "state", e.g. "TX"). When availability returns checked=false it could not be checked: available is null (unknown) for every area code; confirm one with preview_update. To buy numbers use tenant_ai_sinch_numbers_update.
tenant_ai_sinch_numbers_update
ChatGPTBuy Telephony Sinch phone numbers for the active tenant. Inbound numbers are saved to the tenant's phone numbers and billed per number; outbound numbers are not billed per number. One order is one area code and 1 to 10 numbers. Actions: preview_update (checks the order, buys nothing), apply_update (buys it), finish_setup (retries the setup after an apply_update that returned setup_complete=false, for inbound and outbound alike, so it takes no number_type, and returns saved_count; never buys). apply_update requires confirmation="APPLY_TENANT_AI_SINCH_NUMBERS_UPDATE". When apply_update returns setup_complete=false the numbers were bought: run finish_setup, not another apply_update. An error purchase_uncertain means numbers may have been rented: for inbound, run finish_setup (it saves any rented inbound numbers) and check the tenant's phone numbers; for outbound, list them with tenant_ai_sinch_numbers. Do not buy again until that check is done. Check area-code availability with tenant_ai_sinch_numbers. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_sms_agents
ChatGPTList and read the active tenant's SMS agents (the AI assistants that answer inbound SMS). Actions: list, read (by "id"). To create, change or delete an agent use tenant_ai_sms_agents_update.
tenant_ai_sms_agents_update
ChatGPTCreate, change or delete the active tenant's SMS agents. Actions: preview_update, apply_update, delete. apply_update with no "id" creates a new agent (name is required); with an "id" it updates that agent. An agent must have status="active" to be set as default, and only one agent per tenant can be default. apply_update and delete require confirmation="APPLY_TENANT_AI_SMS_AGENTS_UPDATE". List and read agents with tenant_ai_sms_agents. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_tools_configuration
ChatGPTList and read the tenant AI tool overrides (end_call, webhook_dispatcher, transfer_call, send_sms, credit_score). Actions: list, read (one tool, by "tool_name"). Note: a tool row is inert unless the tool name appears in the voice prompt text, and transfer_call additionally requires use_tool=true before the live pipeline injects it. To change an override use tenant_ai_tools_configuration_update.
tenant_ai_tools_configuration_update
ChatGPTChange the tenant AI tool overrides (end_call, webhook_dispatcher, transfer_call, send_sms, credit_score). Actions: preview_update, apply_update, each targeting one tool via "tool_name". apply_update requires confirmation="APPLY_TENANT_AI_TOOLS_CONFIGURATION_UPDATE". Note: a tool row is inert unless the tool name appears in the voice prompt text (PromptConfigService::extractTools scans the prompt), and transfer_call additionally requires use_tool=true before the live pipeline injects it. credit_score takes no parameters. List and read the overrides with tenant_ai_tools_configuration. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_voice_prompt
ChatGPTRead the active tenant's AI voice assistant configuration (prompts, opening messages, tuning, and LLM/voice/STT model selection, with the catalog of valid model/voice/STT ids) and its saved prompt library. Actions: read, list_prompts. To change the configuration or the saved prompts use tenant_ai_voice_prompt_update.
tenant_ai_voice_prompt_update
ChatGPTChange the active tenant's AI voice assistant configuration (prompts, opening messages, tuning, and LLM/voice/STT model selection) and manage its saved prompt library. Actions: preview_update, apply_update, save_prompt, delete_prompt. apply_update and delete_prompt require confirmation="APPLY_TENANT_AI_VOICE_PROMPT_UPDATE". Valid model/voice/STT ids come from tenant_ai_voice_prompt (action read). A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_web_call
ChatGPTRead the tenant Web Calling settings (enabled, allowed_domains, intro_id, consent_text) and the derived public token, widget URL, embed snippet, resolved consent text and widget title. Action: read. Note: an empty allowed_domains list denies EVERY embed - it does not mean "allow all". This tool cannot place a web call; use tenant_ai_call_context_preview with call_type="web". To change the settings or rotate the token use tenant_ai_web_call_update.
tenant_ai_web_call_update
ChatGPTChange the tenant Web Calling settings (enabled, allowed_domains, intro_id, consent_text) or rotate the public widget token. Actions: preview_update, apply_update, regenerate_token. apply_update requires confirmation="APPLY_TENANT_AI_WEB_CALL_UPDATE". regenerate_token requires confirmation="REGENERATE_TENANT_AI_WEB_CALL_TOKEN" and invalidates every snippet already deployed on the customer's sites. Note: an empty allowed_domains list denies EVERY embed - it does not mean "allow all". Read the current settings and snippet with tenant_ai_web_call. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.
tenant_ai_webhooks
ChatGPTList and read the active tenant's outbound webhooks (URL, auth, headers, payload, triggers). Secret credentials come back masked. Actions: list, read (by "id"). To create, change or delete a webhook use tenant_ai_webhooks_update.
tenant_ai_webhooks_update
ChatGPTCreate, change or delete the active tenant's outbound webhooks. Actions: preview_update, apply_update, delete. apply_update with no "id" creates a webhook; with an "id" it updates that webhook. Credentials (passwords, tokens, auth header values, credential-like custom headers) are never entered here: they are set in the BiglySales app, and existing ones are kept unchanged (omit them or resend "***"). apply_update and delete require confirmation="APPLY_TENANT_AI_WEBHOOKS_UPDATE". List and read webhooks with tenant_ai_webhooks. A change is made in two calls to this tool: preview_update returns the diff, then apply_update with the same values and the confirmation string applies it; apply_update is refused without a recent preview_update of exactly those values. The onboarding steps and the questions to settle with the user before the first change are in tenant_ai_onboarding_guide.