compelling_account_timeline
ChatGPTReturn the history of one company account, newest first: CRM deal stages, meetings, statements, messages, replies, outcomes and research answers, each with source and date. Use this when the user asks what happened with an account or prepares a follow-up. Statements come from meetings and the team's chats; outcomes include replies, booked meetings and won or lost deals. Read-only, no credits.
compelling_add_insight
ChatGPTCreate a new insight column (research question) on a list without researching it. Use this when the user wants a new column, such as a custom question or a standard field like revenue, industry, employees, email or LinkedIn profile. It creates the column only and consumes no credits; values are researched with compelling_run_insight or written with compelling_ingest_records. Custom answer types (boolean, singleChoice, multipleChoice, integer, entity, url, address, date, phoneNumber, jobtitle) require question and header_name, plus choice_options for single and multiple choice. Default answer types (for accounts: revenue, industry, employees; for contacts: email, personalLinkedin) need only answer_type.
compelling_add_to_blacklist
ChatGPTExclude company domains from the workspace's lists. Use this when the user asks to hide companies, such as existing customers or competitors, from lists. Pass domains only. Domains already excluded are reported as skipped. Excluded companies no longer appear in list views. No credits.
compelling_apply_template
ChatGPTAdd the insight columns of an enrichment template (saved or built-in) to a list. Use this when the user asks to apply a template to a list. Each template column becomes a new, empty column on the list; columns the list already has, and built-in columns the list has archived, are skipped and reported. It starts no research and consumes no credits; values are researched per column with compelling_run_insight. Lists that an integration researches automatically for every new record are refused, because new columns there consume credits for each future record; the user applies the template in the Compelling app instead. The template id comes from compelling_list_templates, the list id from compelling_list_lists.
compelling_archive_routine
ChatGPTArchive one of elli's routines. Final: it never runs again and cannot be restored; its past runs stay. A run that is going is stopped. Use this only when the user wants a routine gone for good; to stop it for a while, pause it with compelling_update_routine. routine_id comes from compelling_list_routines. Only the routine's owner or a workspace admin may archive it.
compelling_contacts_run_status
ChatGPTReturn the status of a contact-finding run started by compelling_find_contacts. Use this when the user asks about progress of a contact search. By default it waits up to 25 seconds for the run to finish; wait_seconds=0 returns the current state immediately. Counters (inserted, skipped, processed) can keep changing for a few seconds after status='completed'; the saved contacts themselves are read with compelling_read_list(record_type='contacts'). Read-only, no credits.
compelling_copy_rows
ChatGPTAdd companies and/or contacts the workspace already has to another list. A contact brings its company. They keep their place on other lists; the target list's columns start empty for them until researched. No credits.
compelling_create_list
ChatGPTCreate a new, empty list in the workspace and return its id. Use this when the user asks for a new list, or when a requested company search, contact search or new column needs a list that does not exist yet. No credits.
compelling_create_outreach_campaign
ChatGPTCreate a paused outreach campaign in a connected outreach tool with an e-mail sequence and the list's contacts as leads. Use this when the user asks to prepare a campaign for a list in Instantly, Salesforge or HeyReach. The campaign is created paused and never starts sending; the user reviews and starts it in the outreach tool. For HeyReach the contacts are added to a new lead list (or a chosen paused campaign) and the e-mail steps are not transferred; the result states this. Requires the outreach tool to be connected (link via compelling_get_integration_link). Contacts without an e-mail address (HeyReach: without a LinkedIn profile URL) are skipped and counted. No credits. Returns the campaign id, numbers of leads added and skipped, and a link to the campaign.
compelling_credit_usage
ChatGPTReturn the workspace's credit spending over a time window and its remaining balance. Use this when the user asks how many credits were spent, on what, or how fast. Returns totals, a breakdown by usage type (company research, contact finding, company search criteria, ranking) and a daily series. Read-only, no credits. Costs of a future action come from compelling_estimate_credits.
compelling_delete_insight
ChatGPTDelete an insight column from its list, as the app's delete does. Its filter and sort rules are removed from the list's views. No credits.
compelling_delete_view
ChatGPTDelete a named view of a list. The default view cannot be deleted; records and research stay. No credits.
compelling_estimate_credits
ChatGPTEstimate the maximum credits one action would consume and return the available balance, without starting it. Use this when the user asks what an action would cost or whether the balance is enough. Supported actions: find_contacts (2 credits per contact), run_insight (depends on answer type and record type), find_companies (5 base plus one per standard or optional criterion, per company) and rank_list (1 credit per account). Actual spend never exceeds the estimate. Read-only: it starts nothing and charges nothing.
compelling_find_companies
ChatGPTStart a search that finds new companies matching the user's market and criteria and saves them to a list. Use this when the user asks to build a new list of target companies, for example machinery manufacturers in Germany using SAP, or medtech companies in Austria and Switzerland with ISO 13485 certification. Keep the geography and segment as requested; DACH means Germany, Austria and Switzerland. Required criteria decide membership: a company is added only if it meets all of them. Optional criteria are researched, scored and billed, but do not decide membership. Each criterion becomes a column on the list. The search runs asynchronously, uses external sources and consumes credits. The requested company count is a target, not a guarantee. Progress is available from compelling_find_companies_status; saved companies from compelling_read_list(record_type='accounts'). Companies already stored are found with compelling_search_records. Cost estimates without starting come from compelling_estimate_credits(action='find_companies').
compelling_find_companies_status
ChatGPTReturn the status of a company search started by compelling_find_companies. Use this when the user asks about progress of a company search. Waits until the search ends (complete, exhausted, cancelled or prompt_ambiguous) or wait_seconds elapses. Found companies are read with compelling_read_list. Read-only, no credits.
compelling_find_contacts
ChatGPTStart a job that finds business contacts by role (names and job titles) at companies in a list. Use this when the user asks for decision-makers or other contacts at companies in a list. roles carries the requested titles or persona as the user stated them, including German titles; account_ids and limit restrict the companies, and per_title_limit caps contacts per company and role. The job runs asynchronously, uses external sources, saves matched contacts to the workspace and the list, and consumes credits (2 per contact found). Returns a run_id and status, plus progress counts if wait_seconds is set. Progress is available from compelling_contacts_run_status; saved contacts from compelling_read_list(record_type='contacts'). It finds names and job titles only: email addresses, phone numbers and LinkedIn profiles are separate insight research. It does not contact anyone. Cost estimates without starting a job come from compelling_estimate_credits(action='find_contacts').
compelling_get_account_knowledge
ChatGPTReturn a summary of what the workspace knows about one company: saved research, fit with the target profile, missing information and suggested next research. Use this when the user asks what is known about a company, whether it fits their target profile, or what to research next. Identify the company by account id or domain. Suggestions come from fixed rules about missing values; if the company fails the fit check, no further paid research is suggested. Read-only, no credits.
compelling_get_company_model
ChatGPTReturn the workspace's Company Model as Markdown: the user's own offer, customer and partner segments, and related context. The Company Model describes the user's company: offer, customer and partner segments with buying center, pains, use cases, evidence and competitors (unconfirmed items are marked), recommended ideal-customer searches per segment, recent company searches and the most recently edited lists. Use this when the user asks about their target segments or ideal customer profile, or asks for a new list, search criteria or outreach angle based on their business. Read-only, no credits.
compelling_get_import_link
ChatGPTReturn links to the Compelling web app where the user can import a CSV file or sync a list from a connected app. Use this when the user wants to bring data into Compelling from a file or from HubSpot, Pipedrive, Google Sheets or similar. It does not import or upload anything itself: the user opens the link and completes the import in the browser. An app that is not connected yet needs compelling_get_integration_link first. Read-only.
compelling_get_integration_link
ChatGPTReturn a link to the Compelling web app where the user can connect one integration (for example HubSpot, Pipedrive or Google Sheets). Use this when the user asks to connect an app. It does not connect anything itself and does not handle credentials: the signed-in user opens the link in a browser and completes the connection there. Read-only.
compelling_get_workspace
ChatGPTReturn the name and domain of the Compelling workspace this connection uses. Use this when the user asks which workspace or company account is connected. The workspace is fixed by the authenticated connection; the result contains no internal identifiers. Read-only, no credits.
compelling_ingest_records
ChatGPTWrite companies, contacts and known column values supplied by the user directly into a list. Use this when the user provides data to store in Compelling, such as companies, contacts or values for existing columns. No research and no credits. Existing answers for the supplied question_ids are overwritten, so the records and columns to replace need the user's confirmation first. Columns must exist (compelling_add_insight creates them).
compelling_insight_status
ChatGPTReturn the progress of research on an insight column started by compelling_run_insight. Use this when the user asks whether research has finished. Waits up to wait_seconds until every targeted answer is finished (done=true). Counts cover the whole column unless account_ids or contact_ids narrow them to the records of one run, and can lag a few seconds; the researched values are read with compelling_read_list. Read-only, no credits.
compelling_list_active_work
ChatGPTList the workspace's running jobs (company searches, contact searches, research and exports) with their list and run ids. Use this when the user asks what is currently running, or when a run_id or list_id of a running job is needed. Progress of each job is available from its status tool (compelling_find_companies_status, compelling_contacts_run_status, compelling_insight_status). Read-only, no credits.
compelling_list_blacklist
ChatGPTList the company domains excluded from the workspace's lists. Use this when the user asks which companies are excluded or hidden. Excluded domains do not appear in list views. Read-only, no credits.
compelling_list_favorite_insights
ChatGPTList the workspace's favorite insights: id, column label, question, answer type, options. A favorite insight is a custom insight column (a research question) that someone in the workspace starred so it can be reused on other lists; favorites are shared by the whole workspace. Use this when the user asks which favorites exist or wants to put a favorite onto a list: pass its favorite_id to compelling_add_insight. Read-only, no credits.
compelling_list_from_view
ChatGPTCreate a new list from the rows a view or filter shows on an existing list, with the same columns and their answers. A filter is required (saved on the view or passed here). No credits.
compelling_list_insights
ChatGPTList the insight columns of a list: question_id, header and answer type. An insight column is a research question on a list (for example "Uses SAP?" or employee count) whose answers are stored per record. Use this when the user refers to a column by name and its question_id is needed for compelling_run_insight, compelling_read_list (fields, filters, sort_by) or compelling_rank_list. Also lists the system column addedAt, the date a record was added to the list, which compelling_read_list filters accept. Read-only, no credits.
compelling_list_integrations
ChatGPTList the CRM, data and outreach integrations available to the workspace, with connection status and app links. Use this when the user asks which tools are or can be connected. Read-only. Connecting an app requires the user to finish sign-in in the Compelling web app; compelling_get_integration_link returns the link for that.
compelling_list_lists
ChatGPTList the workspace's lists with id, name and account and contact counts. Use this when the user asks which lists exist or names a list whose id is needed. Sorted by last edit, newest first; supports pagination. Read-only: it does not change lists or records and consumes no credits.
compelling_list_outreach_campaigns
ChatGPTList the campaigns (sequences) in a connected outreach tool with their id, name and status. Use this when the user asks which campaigns exist in Instantly, Salesforge or HeyReach, or wants to load a list's contacts into one of their existing campaigns. Read-only: it changes nothing in the outreach tool and consumes no credits. Status is sending, paused, draft, finished or other; only campaigns marked as taking leads can be passed as target_campaign_id to compelling_create_outreach_campaign. Requires the outreach tool to be connected (link via compelling_get_integration_link).
compelling_list_routines
ChatGPTList everything under Routines in this workspace, exactly what the Routines page shows: elli routines, scheduled Insight columns and elli dailies, soonest next run first. Each with its id, kind, list, status (active, paused, draft), schedule and next run in time_zone, the last run, the estimated credits a month, whether the user may change it, and the link to its page. Use it when the user asks which routines exist or about one of them, before creating a routine that may already exist, and for the routine_id of compelling_update_routine and compelling_archive_routine. Never guess a routine's link.
compelling_list_runs
ChatGPTList what runs and ran in the workspace: company searches, contact searches, research, imports, exports and elli routine runs, with status, progress, failure cause and a link. Use this when the user asks what is running, what ran recently, whether a job finished or why it failed, and before preparing a job for a list to see whether one already runs there. Active jobs come first, then finished ones newest first. A failure is a cause category (no_credits, integration_disconnected, timed_out, not_in_plan, column_deleted, no_answer, over_estimate, unclear_prompt) or none. Times are UTC. Read-only, no credits.
compelling_list_setup
ChatGPTHow a list is set up: its saved views with their filters and sorting, and the sources that import into it (OMR Reviews, SalesViewer, HubSpot, Pipedrive, Dynamics, Attio, Dealfront, Granola, Kickscale, Google Sheets) with whether each imports and researches new records automatically and whether it can import older records. The default view is what the list shows; new_accounts_researched says whether its filter picks the new accounts that get researched or every new account is. Read-only, no credits.
compelling_list_templates
ChatGPTList the enrichment templates with id, name and columns: the workspace's saved ones, then the built-in ones every workspace has. An enrichment template is a named set of insight columns (research questions for companies and contacts), so the same columns can be added to other lists. Saved templates come from the workspace's own lists, sorted by name; built-in ones (e.g. "Hiring signals", "ICP fit check") have a slug as id. Use this when the user names a template or asks which templates exist; the id is needed for compelling_apply_template. Read-only, no credits.
compelling_preview_crm_sync
ChatGPTCheck, without writing anything, which of a list's companies a CRM sync would update in the CRM and which it would create as new accounts. Use this before proposing or starting compelling_sync_list_to_crm, and when the person asks whether a sync creates duplicates or how existing accounts are found. The sync looks each company up the way the export does: a link from an earlier sync, then the website (however the CRM spells it), then the name; a company with neither match is created. Available for Dynamics 365; other CRMs answer that no preview exists. Reads the CRM, consumes no credits.
compelling_rank_list
ChatGPTStart scoring a list's companies from 0 to 100 against a weighted rubric of existing columns. Use this when the user asks to prioritize companies in a list by fit or decide which to approach first. The rubric names existing column IDs with weights and point mappings for their values (for example employee count, technology or certification); explain the rubric and weights to the user. A score reflects fit with the rubric, not purchase intent; missing values count as unknown. Runs asynchronously, saves a ranking column and consumes 1 credit per account. Companies only; it does not rank contacts or research missing values. Progress is available from compelling_ranking_status; sorted results from compelling_read_list with sort_by set to the ranking column's question_id. Cost estimates without starting come from compelling_estimate_credits(action='rank_list').
compelling_ranking_status
ChatGPTReturn the progress of a list ranking started by compelling_rank_list. Use this when the user asks whether scoring has finished. Can wait until scoring completes. response_format='json' returns status, scored_count, total_accounts and last_generated (when the last scoring pass finished). During a re-rank, status shows 'complete' from the previous pass until the new pass finishes; compare last_generated with the start of the re-rank. Read-only, no credits.
compelling_read_list
ChatGPTRead the saved companies or contacts of one list, with optional column selection, filters and sorting. Use this when the user wants to view, filter or sort records and their saved insight values (for example Industry or Employees). fields selects columns by header; record IDs and display names are always included. Filters and sorting apply to this response only; sorting by a question_id is available for accounts (column IDs come from compelling_list_insights). Read-only: it does not change the list or its saved filters, does not research missing values, and consumes no credits. Values that were never researched are returned empty.
compelling_read_meeting
ChatGPTRead the original notes and transcript of one meeting the workspace imported from Granola or Kickscale. Use this when the user needs an exact detail from a meeting (a number, a wording, who said what) that the account timeline does not contain. The content is fetched from the meeting tool at request time and not stored. Only meetings this workspace imported are available. Long transcripts are paged; query returns only matching passages. Read-only, no credits.
compelling_remove_from_blacklist
ChatGPTTake domains off the workspace blacklist. Their accounts show in lists again. Companies a past import skipped because of the blacklist do not come back on their own; they need a new import. Domains not on the blacklist are reported as skipped. No credits.
compelling_remove_rows
ChatGPTTake companies and/or contacts off a list. They stay in the workspace with their research; a company's contacts leave the list with it. No credits.
compelling_run_insight
ChatGPTStart research that fills an existing insight column for selected companies or contacts in a list. Use this when the user asks for new research on records in a list, such as SAP usage, ISO 13485 certification, hiring activity, revenue or employee count for companies, or business email and LinkedIn profile for contacts. The question configured on the column is researched; keep any research period the user requested. Runs asynchronously, uses external sources, consumes credits and saves answers to the list. Rerunning replaces existing answers in that column, including generated message text, so the records, column and any replacement need the user's confirmation first. It does not send messages. Requires account_ids or contact_ids matching record_type, or a limit; question_id comes from compelling_list_insights. Progress is available from compelling_insight_status; saved answers from compelling_read_list. Answers can be inconclusive when no evidence is found. Cost estimates without starting research come from compelling_estimate_credits(action='run_insight').
compelling_save_view
ChatGPTSave filters, sorting, the tab and shown columns on a list: change an existing view or create a named one. A view belongs to the list, not to a person: a change shows for everyone in the workspace. The default view also steers the list. Its filter decides, for lists fed by a source with auto-research on, which new records get researched (and use credits); its shown columns are the ones researched for them, so showing a column there costs credits per new record and hiding one stops that. Its shown columns, filter and sort make the CSV/Excel export. Its tab cannot change here (error default_view_tab): save a new named view with that tab. A view with an active export sync sends the rows its tab, filter and sort select, so changing them changes what goes to that CRM. The approval card names each of these effects. Nothing is researched now and no credits are used. Returns the view and its app link.
compelling_search_records
ChatGPTSearch existing companies or business contacts in the workspace by name or domain. Use this when the user wants to find a company or contact already stored in Compelling. Returns record IDs, names, company domains or contact job titles and employers, and list memberships; paginated with limit and offset. Read-only: it searches stored data only, does not discover new companies, does not use external sources, and consumes no credits.
compelling_set_favorite_insight
ChatGPTMark a custom insight column as a favorite insight of the workspace, or remove it. A favorite insight can be put onto other lists with compelling_add_insight(favorite_id=...) and is visible to everyone in the workspace. Only custom columns (the workspace's own research questions) can be favorites, not built-in ones such as email or revenue. Removing a favorite keeps the column and its answers on its list. Consumes no credits.
compelling_set_list_automation
ChatGPTChange a list's automations: pause or resume the automatic import and/or automatic research of new records, and import a source's older records into the list. With auto-research on, every new record the source brings, older imported ones included, is researched with the list's columns and uses credits; the default view's filter limits which ones. With it off nothing is researched. Companies already on the list are skipped.
compelling_stop_run
ChatGPTStop a running contact search, or pause a running company search. Work already started finishes; nothing more is started. Insight research cannot be stopped once started.
compelling_sync_list_to_crm
ChatGPTSync a list's companies and contacts to a connected CRM (HubSpot, Pipedrive, Attio, Dynamics 365 or monday.com) using the workspace's field mapping. Use this when the user asks to push a list to their CRM. Creates or updates records in the CRM according to the field mapping configured in Compelling. Requires the CRM to be connected and mapped; otherwise it returns an error with a link for the user. Runs in the background, consumes no credits, and returns the sync run id and status.
compelling_update_insight
ChatGPTChange an existing insight column: its label, visibility, question, answer type or options. Label and visibility change in place on any column. Question, answer type and options change in place while the column has not been researched yet. Once research has run on it, the old column keeps its answers and the change creates a new column next to it with the changed question; the result names that new question_id. Nothing is researched and no credits are used; research the column with compelling_run_insight.
compelling_update_list
ChatGPTRename a list, archive it (the app's delete: it disappears from the lists) or restore an archived one. Records and research stay in the workspace. No credits, but a restored list an integration researches is researched again for every new record, which costs credits.
compelling_update_my_settings
ChatGPTChange the elli settings of the person you act for: when their daily comes, whether memory is paused, and where proposals reach them (Slack, email). Pass only what they asked to change. Only their own settings; workspace defaults are for admins in the app. They see every proposal in the app anyway. No credits.
compelling_update_routine
ChatGPTPause, resume, rename or reschedule one of elli's routines. Use this when the user wants a routine paused or resumed ("pause the Monday routine"), renamed or run at another time. Pass only what changes; routine_id comes from compelling_list_routines. A new schedule is a new version of the routine, and from then on it runs as the user who confirms it; a routine runs once a day at most. Only the routine's owner or a workspace admin may change it. Scheduled Insight columns change with compelling_schedule_insight, elli dailies on their page.
compelling_update_workspace
ChatGPTChange workspace settings: name, domain, value proposition, language or premium phone numbers. Use this when the user asks to change one of these settings. Pass only the fields to change. The value proposition is replaced as a whole, so pass the complete new text (the current text is in compelling_get_company_model). It is used for search suggestions, research and generated messages, and the app regenerates its example searches from it. No credits.