add_flow_node
ChatGPTAdd one step after an existing step (after = a node id from get_flow; on a condition also pass branch). It takes over whatever followed there, like inserting into a line in the builder. An email brings its own 5-day wait; at most 3 emails per flow. Never sends or writes anything.
create_campaign
ChatGPTCreate a campaign from a template (template_id from list_flow_templates) or from a sentence (text, e.g. "test units for the cordless vacuum lists"). Picks the best pages AI answers cite (Sources), saves flow, pitch and trigger. Writes no emails and costs nothing: call draft_emails for that. Returns campaign_id and url; pass campaign_id to every following call.
delete_flow_nodes
ChatGPTDelete steps; what came before links to what came after. All or nothing; the start cannot be deleted. Only delete steps the user asked to remove.
disconnect_flow_nodes
ChatGPTRemove the edge from → to. Both steps stay; the target is unconnected (never runs) until connected again. All pairs or none.
dismiss_request
ChatGPTMove a request to Dismissed (reopen = false), or back to Open (reopen = true).
draft_status
ChatGPTHow far draft_emails got: pages, drafted, pages with a contact, pages still without a draft. Poll it every poll_after_seconds until all_completed.
duplicate_flow_nodes
ChatGPTCopy steps with their settings. Copies of two copied steps stay linked; with copy_incoming_edges each copy is also wired to its original's source where that source has a free output.
get_answer_content
ChatGPTRead the actual AI answers for one prompt on one day: the answer text, all brands named in the answer with their sentiment, and every source the model cited or saw. Defaults to the most recent day with answers; pass date (YYYY-MM-DD, see get_prompt_details availableDates) for a specific day and model to narrow to one AI model. response_format: "concise" (default) truncates long answer texts to save tokens; use "detailed" for the full text.
get_competitor_gap_analysis
ChatGPTGap analysis vs one competitor (same numbers as the Gap Analysis tab on the competitor detail page). Returns aggregate counts over distinct (prompt, model, day) answers — onlyThem (competitor mentioned, you are not = your losses), both (head-to-head), onlyYou (your wins) — plus a sample of the concrete prompts behind them. Use gapType to choose which prompt list you want: "competitor_only" (default — where you lose), "user_only" (where you win) or "both". Get the competitorId from get_competitor_ranking (field id).
get_competitor_h2h
ChatGPTHead-to-head comparison of your brand vs one competitor over answers that mention BOTH brands: youFirst/themFirst/tied (who is named first, based on ordinal mention position), average mention order for each side, and LLM-judged comparative claims (claims.you/claims.them = claims favoring each side). Get the competitorId from get_competitor_ranking (field id).
get_competitor_ranking
ChatGPTGet the competitor visibility ranking for a project: which brands are mentioned most in AI answers over the timeframe. visibilityRatePercent is the share of analyzed AI answers where the brand is visible — named in the answer text OR its domain cited as a source — the dashboard "Visibility" column; mentionRatePercent is the share of answers naming the brand in the text ("Mention Rate" column); citationRatePercent is the share of answers citing that brand's domain as a source ("Citation Rate" column). All three are comparable with the same-named fields of get_visibility_metrics. avgMentionOrder is the ordinal "Position" KPI (average rank among all brands named in an answer, 1 = named first — report this when asked about position); mentionDepthPercent is how deep in the answer text the mention appears (NOT a rank). Each competitor row includes an id — pass it as competitorId to get_competitor_gap_analysis or get_competitor_h2h to drill deeper into one competitor. totalCitations per competitor is ANSWER-LEVEL: the number of unique AI answers citing that brand's domains in the timeframe — the same unit as the dashboard "Citations" KPI and get_visibility_metrics ownDomainCitations.total (which covers only the own brand's exact domain).
get_entity_tree
ChatGPTThe project's entity tree: how detected names are grouped into companies → sub-brands → products → ingredients, which levels exist, the own brand's subtree (your company, product brands, ingredient names) and the competitors with their sub-entities. Call it before using entityLevel on get_competitor_ranking, get_visibility_metrics, get_visibility_timeseries or get_sentiment_overview, and whenever the user asks about a product, ingredient or subsidiary rather than the company. ownOptions lists which own entities can be "you" per level (pass one as ownEntity). stage "live" means the tree is switched on for the project; otherwise every tool reports the company-level grouping and entityLevel is ignored.
get_flow
ChatGPTRead a campaign's flow: steps with their next steps, open_outputs (where add_flow_node can attach a step), unconnected steps (never run until connected) and a node_url per step. Pass node_ids or label_contains to read just part of a large flow. Read it before changing a flow.
get_flow_node
ChatGPTOne step with all its settings, the steps before and after it, and its node_url.
get_prompt_details
ChatGPTDrill into ONE tracked prompt: prompt metadata, aggregated metrics, per-model metrics, and the answer history — one row per AI answer (newest first) with date, model, whether your brand was mentioned, mention depth, sentiment and the brands named in that answer. Get the promptId from list_prompts (field id). To read the full text of a specific answer, use get_answer_content with the same promptId plus a date from availableDates.
get_query_fanouts
ChatGPTGet the AI query fan-outs (related sub-queries that AI models generated while answering your tracked prompts), aggregated and ranked by frequency. Each fan-out includes total occurrences, distinct prompts that triggered it, models, and last-seen timestamp.
get_request
ChatGPTOne request: the mail, why it ranks where it ranks, and the drafted answer. Answers are sent only by a person from the Inbox in Finseo (review_url).
get_sentiment_overview
ChatGPTSentiment and Share-of-Voice overview for a project (same data as the dashboard Sentiment page): uncapped Share of Voice (own mentions vs ALL detected brands, and vs your tracked list — both denominators are explicit), a per-brand sentiment leaderboard (score 0-100, positive/neutral/negative distribution, win rate, strongest/weakest aspect, per-aspect scores) and the most negative LLM-judged statements with their aspect and date.
get_tags
ChatGPTList all tags for a project with prompt counts
get_task_details
ChatGPTRead ONE optimization task in full: description, step-by-step action plan, the evidence (data points and affected prompts/sources) that triggered it, and — for content tasks — a structured content plan (target page, template, prompts to cover). Get the taskId from list_tasks (field id).
get_top_sources
ChatGPTGet the most cited sources in AI responses for a project
get_visibility_metrics
ChatGPTGet brand visibility KPIs for a project, aggregated over the selected timeframe and compared with the previous period of equal length. IMPORTANT: visibilityRatePercent (share of analyzed AI answers where the brand is visible: named in the answer text OR its own domain cited as a source) is the "Visibility" KPI shown in the Finseo dashboard — use this when reporting visibility. mentionRatePercent = share of answers naming the brand in the text; citationRatePercent = share of answers citing the own domain. promptCoveragePercent (share of tracked prompts with at least one mention) is a different, usually much higher number.
get_visibility_timeseries
ChatGPTGet per-day visibility for a project over the timeframe — one row per day with visibilityPercent (share of that day's AI answers where the brand is visible: named in the text OR own domain cited as a source; same as the dashboard chart), mentionRatePercent (named in the text), citationRatePercent (own domain cited), brand mentions, own-domain cited answers, answers analyzed, avgMentionOrder (ordinal "Position" KPI: average rank among brands named in an answer, 1 = first), mentionDepthPercent (how deep in the answer text the mention appears — NOT a rank) and sentiment. Use this for questions like "how did my visibility change day by day".
list_campaigns
ChatGPTThe project's outreach campaigns (each one is a flow) with status, trigger and stats, plus a url that opens the flow builder.
list_flow_node_types
ChatGPTThe step kinds an outreach flow can hold (email, wait, if/else with its branches, task, AI instructions, LinkedIn, end), with what each does and which fields it takes. Call this before add_flow_node rather than guessing a kind.
list_flow_templates
ChatGPTReady-made outreach flows (built-in playbooks and the team's own templates) to start a campaign from. Pass a template_id from here to create_campaign.
list_projects
ChatGPTList all Finseo projects (tracked brands/domains) for the authenticated user
list_prompts
ChatGPTList tracked AI prompts with summary metrics (isVisible = named in the text OR own domain cited in at least one answer, mentions and mentionRatePercent, own-domain citations and citationRatePercent, mention depth, sentiment, top competitors). Returns clean prompt summaries without raw response data. Each prompt row includes an id — pass it as promptId to get_prompt_details (per-answer history) or get_answer_content (full answer text) to drill deeper. Paginated: use offset together with hasMore in the response to fetch further pages.
list_requests
ChatGPTRequests that reached the project's press inbox (statements, test units, guest articles), best potential for AI visibility first (potential 0–100 from AI citations of the outlet, whether AI crawlers can read its pages, on-page quality and topic fit; null while it is being checked; score = citations, topic and deadline only), with whether an answer is drafted.
list_tasks
ChatGPTList the AI-generated optimization tasks for a project (the Tasks page in the Finseo dashboard) — concrete, prioritized to-dos derived from the project's visibility data (e.g. fix crawl errors, answer high-potential community threads, close competitor gaps, rework weak content). Rows are sorted by priorityScore (impact weighted against effort; higher = do first) and stay compact — pass a row's id to get_task_details for the full description, step-by-step plan and evidence. Statuses: open, in_progress, done, dismissed. Categories: technical, visibility, content, offsite, reputation, competitor, setup.
get_competitor_ranking
Claudeget_query_fanouts
Claudeget_tags
Claudeget_top_sources
Claudeget_visibility_metrics
Claudeget_visibility_timeseries
Claudelist_projects
Claudelist_prompts
Claude