Iterate

Returns aggregate metrics over feedback (server-side counts, never raw responses) so you can answer questions like top themes, sentiment split, and which survey is most negative without reading rows. All inputs optional: survey_id (omit to span every survey), question_id, since/until (RFC3339), group_by (none, survey, week, month), metrics (any of volume, sentiment, topics, scores; defaults to volume, sentiment, topics), top_n (topics per group, default 10). Use list_surveys to get a survey_id; use get_responses only when you need the actual text.

aggregate_feedback

Creates one safe, non-live survey draft after the customer has explicitly confirmed the complete proposal. Before calling: ask what the customer is trying to learn, ask only essential audience and delivery follow-ups, use get_survey_creation_context, propose the 3–5 best questions, and show the full survey, including advanced options, for confirmation. When an existing survey appears relevant, use list_surveys and get_survey_questions to understand the account's terminology and avoid duplication; do not copy it verbatim. Supported question types are freeform, single_choice, multiple_choice, rating, nps, and csat. Questions should be neutral, focused on one idea each, varied when useful, and include an open-ended question when it will add insight. All questions default to optional and may have a subheader or be required. Choice questions support Other with a custom label and randomization; multiple choice supports selection limits; freeform supports a character limit. Connected integration delivery methods may be selected when returned by get_survey_creation_context, but targeting, recipients, and campaign configuration happen later in Iterate. This tool never sends email, contacts respondents, displays a survey, or publishes an integration campaign. Call it once after confirmation; if the outcome is unclear, use list_surveys to check before trying again. If validation requires a material revision, show the revised draft and obtain confirmation again before retrying.

create_survey

Returns the authenticated Iterate user's email and company name. Use it to confirm which account is connected.

whoami

Returns the writable projects and the eligible and unavailable delivery methods for the connected Iterate account, including email and in-app variants for connected integrations. Call this before proposing a survey draft. Ask what the customer is trying to learn first, ask only essential follow-up questions, and normally propose 3–5 varied, neutral questions. Targeting, recipients, and external campaign configuration happen later in Iterate.

get_survey_creation_context

Returns a survey's name, delivery and state flags, and ordered top-level question IDs, types, respondent-facing prompts, choices with stored values, and supported question settings. Get a survey_id from list_surveys. State flags do not prove a survey has been shared or delivered. Use this to review a created draft or understand an existing survey's terminology and avoid duplicating questions; do not copy it verbatim.

get_survey_questions

Returns recent feedback responses for one survey (newest first), each with its question, answer, sentiment, and topics. Get a survey_id from list_surveys. Use limit (default 25, max 100), since (RFC3339), and cursor (next_cursor from a prior call) to page.

get_responses

Lists the surveys on the authenticated user's Iterate account, each with its id, name, type, platform, creation date, and total response count. Pass a survey's id to get_survey_questions to read its questions or get_responses to read its feedback.

list_surveys

Searches feedback responses across surveys (newest first) and returns matching quotes, each with its survey, question, answer, sentiment, and topics. Use it to answer things like what customers are saying about a keyword across every survey. All inputs optional: query (free-text match on the response), survey_id (omit to search all surveys), sentiment (positive/neutral/negative), topic (exact match), since/until (RFC3339), limit (default 25, max 100), cursor (next_cursor from a prior call). For every response in one survey in order, use get_responses instead.

search_responses