MCP App Store
by KeenEthics
Humantic AI icon

Humantic AI MCP Server for ChatGPT

By Humantic AI

Overview

Humantic AI brings buyer intelligence into ChatGPT. Connect your Humantic account to read any prospect's DISC personality from a LinkedIn URL or work email (Agent Pi), generate account-research reports on target companies with buyer-intent signals (Agent Miia), and draft outreach matched to a recipient's personality. Humantic AI is our own product; this connects ChatGPT directly to Humantic's platform. A Humantic subscription is required and usage draws on your existing Humantic credits.

How to connect

  1. ChatGPT: Open the Humantic AI listing in the ChatGPT app directory and choose Connect. ChatGPT listing Humantic AI exposes 14 tools here.

Tools

humantic_content_personalize_v1

ChatGPT
Generate or rewrite sales outreach content matched to the recipient's DISC personality. responsetype selects the mode: personalizedemail (rewrite an email draft — content required), firstpitch (cold first-pitch email — prompt, user_context required, 5-1000 words each), reply / followup (thread continuation — content and earlier_message required). Identify the recipient by email OR linkedin_url (exactly one), or pass personalitytype directly if you already know their DISC type. Works without a pre-existing Humantic profile (upstream resolves one on the fly), but creating the recipient's profile first via humantic_pi_profile_create_v1 gives faster, higher-confidence results. Each generation consumes a content-personalization credit (a separate quota pool from profile credits); identical args within 24h return the cached result without consuming credits — pass force=true to regenerate. The response envelope's _idempotency_hit tells you whether the call was cached.

humantic_miia_report_create_v1

ChatGPT
Start a Humantic MIIA account-intelligence report on a target (buyer) company. REQUIRES buyer_company and analysis_type (moderate / deep — deeper = slower and more credits). Optionally pass seller_company and free-text context to tailor relevance. If this call fails with INVALID_INPUT about unresolvable company domain(s), ask the user to confirm the correct company/domain and retry with buyer_company_domain / seller_company_domain — do NOT retry on your own with reworded names, dropped fields, or guessed domains. Reports are ASYNCHRONOUS: this returns immediately with a create_request_id and _status:"processing". A report is typically either ready within ~2 minutes or takes 15-20 minutes — there is little in between. Poll humantic_miia_report_status_v1 with that create_request_id: check once immediately, once more ~2 minutes later, and if still processing, wait until ~15 minutes have elapsed before checking again (polling every few seconds wastes calls with no new information). Once COMPLETED, call humantic_miia_report_fetch_v1 to retrieve the finished report. Each report consumes a MIIA report credit (a separate quota pool from Pi profile credits); identical args within 24h return the cached create without consuming credits — pass force=true to start a fresh report. Requires the API key to have MIIA access.

humantic_miia_report_fetch_v1

ChatGPT
Retrieve the full MIIA account report by create_request_id (from humantic_miia_report_create_v1). Returns markdown by default (compact, ideal for reading/summarizing); pass format:"html" for the standalone styled render. If the report is still generating, returns _status:"processing" — check humantic_miia_report_status_v1 first, and don't retry faster than ~2 minutes after create, then ~15 minutes after that if still not ready. Completed reports are cached for 30 minutes per-token unless nocache=true.

humantic_miia_report_list_v1

ChatGPT
List all MIIA account reports created under this API key (most recent first), with their create_request_ids and statuses. Use this to discover existing reports before creating a new one or to find a create_request_id to fetch.

humantic_miia_report_refresh_v1

ChatGPT
Use this when an existing account report is out of date and the user wants it researched again - 'refresh the Bayer report', 'this is from March, update it', 'rerun the Wipro account plan with current data'. Regenerates the report with the original buyer, seller and analysis type and returns a NEW create_request_id; the old report stays available under its old id. Pass only create_request_id (from create or list). The report must be finished (ready or failed); the request is rejected while any report for the same buyer and seller is still generating - wait for it. ASYNC like create: tell the user it is running, then use humantic_miia_report_status_v1 with the NEW id - once about 2 minutes later, then not again until about 15 minutes have passed. Uses one MIIA credit. Identical calls within 24 hours return the same result at no cost (_idempotency_hit: true); force=true starts another regeneration. If the user only wants to know what is new at the account, use humantic_signals_per_account_v1 instead - free and instant.

humantic_miia_report_section_get_v1

ChatGPT
Retrieve one or more specific sections of a COMPLETED MIIA report — far cheaper than fetching the whole report. Best practice: call humantic_miia_report_sections_list_v1 first and pass the exact section title(s) you chose. Pass create_request_id, optionally report (full_report = default, or executive_summary), and sections: a single value, or an array, where each value is an exact section title, a 1-based index, OR a topic keyword (e.g. "leadership") — keywords are resolved against section content, so the closest-matching section is returned. Matching is case- and punctuation-insensitive; on a total miss the response lists the available section names so you can retry. Served from the cached report with NO additional upstream call once the report has been fetched.

humantic_miia_report_sections_list_v1

ChatGPT
List the sections of a COMPLETED MIIA report so you can pull only the part you need instead of ingesting the whole report. Pass the create_request_id. Returns each section's title plus a one-line preview, for both the full_report (detailed) and the executive_summary (short). ALWAYS call this FIRST: the user's question (e.g. "how do I reach their leadership?") rarely names a section, so YOU compare the user's intent against these titles + previews, pick the most relevant section(s), and then call humantic_miia_report_section_get_v1 with the exact title(s) you chose. Served from the cached report: once the report has been fetched it makes NO additional upstream call. If the report is still generating, returns _status:"processing".

humantic_miia_report_status_v1

ChatGPT
Lightweight poll for a MIIA report's generation status. Pass the create_request_id from humantic_miia_report_create_v1 (or category + buyer_company_domain + analysis_type). Returns _status:"processing" while the report is PENDING / DATA GENERATED and _status:"ready" once COMPLETED. Recommended cadence: check once immediately after create, once more ~2 minutes later, then — if still processing — wait until ~15 minutes have elapsed before checking again; reports rarely finish in between. Not cached. Use this to poll before calling fetch.

humantic_pi_profile_create_v1

ChatGPT
Create a Humantic Pi profile from EXACTLY ONE identifier (LinkedIn URL OR email OR free-form text). Returns the persisted profile when ready. To enrich with additional text later, use humantic_pi_profile_update_v1. If you have both a LinkedIn URL and an email for the same person and want both to resolve on fetch, call create separately for each — Humantic does not expose an automated linking operation.

humantic_pi_profile_fetch_v1

ChatGPT
Fetch an existing Humantic Pi profile by EXACTLY ONE identifier (profile_id, LinkedIn URL, OR email). Cached for 30 minutes per-token unless nocache=true. Returns NOT_FOUND if the profile doesn't exist. By default returns JSON (format=json). Pass format=html for one resource block: enriched profile as text/html under resource, with metadata / usage_stats / _status / _trace_id / _quota on _meta (same values as the JSON envelope).

humantic_pi_profile_flag_v1

ChatGPT
Record correctness/accuracy/usefulness feedback on an EXISTING Humantic Pi profile. Use iscorrect=false with correctid to mark the profile as the WRONG person (correction). Use isaccurate / isuseful for quality feedback. Synchronous, no long-poll. Requires the profile to already exist; returns NOT_FOUND otherwise.

humantic_pi_profile_update_v1

ChatGPT
Enrich an existing Humantic Pi profile with additional text (bio, résumé excerpt, notes, writing sample) to improve confidence and unlock personality (DISC) analysis when previously insufficient. If no profile exists for the given identifier yet, one is created from this call. Accepts EXACTLY ONE identifier (linkedin_url, email, or profile_id). text is REQUIRED. Blocks until upstream LP returns. Pass force=true to bypass the 24h idempotency cache when you want to actually re-trigger upstream re-analysis (otherwise identical args within 24h return the cached prior response without consuming credits). Note: even on _status=ready, Humantic's downstream re-analysis may queue asynchronously — if confidence/DISC haven't moved, the response envelope's _idempotency_hit field tells you whether the call was cached or actually executed.

humantic_signals_all_accounts_v1

ChatGPT
Use this when the user asks what is happening across their accounts - 'what is new in my accounts this week?', 'which accounts should I reach out to today?', 'show my high-impact signals', 'any trigger events?', a weekly pipeline review. Returns buying signals from every account report under this key in one call, newest first. Each has buyer_company, a headline, why it matters to the seller, impact (High, Medium, Low), a suggested action, source link and event date. No create_request_id needed. ALWAYS narrow the feed: impact when the user wants only what matters, buyer_company when they name an account, and date_range (last_7_days, last_1_month, last_3_months, last_6_months; default last_3_months) or since/until. limit defaults to 200. If truncated is true, call again with next_offset - do not assume you saw everything. Free - no credit. For one account tied to its report, use humantic_signals_per_account_v1.

humantic_signals_all_v1

ChatGPT
Get the consolidated buyer-intent signal feed across EVERY MIIA account report under your API key in a single call — no create_request_id needed. Each signal is annotated with its buyer_company (which report it came from), newest first, and the response also lists all buyer_companies covered. Choose the window with date_range (last_7_days / last_3_months / last_6_months, default last_3_months) or a custom since/until range; cap the count with limit (default 200). Use this for "show me all my signals" or cross-account monitoring; use humantic_signals_list_v1 when you want a single report's signals. This is a free read — it does not consume any credit.

humantic_signals_list_v1

ChatGPT
Get the buyer-intent signals for a COMPLETED MIIA account report — the Bombora-style intent feed (topic surge/strength, impact High/Medium/Low, a suggested action, source link, and event date) for the target (buyer) company. Pass the create_request_id from humantic_miia_report_create_v1 / humantic_miia_report_list_v1; the server resolves it to the report internally (you never pass a report_id). Returns the newest signals first, defaulted to and capped at the last 60 days — narrow the window with since and cap the count with limit. Signals are refreshed on a schedule (typically a few times a week), so the same report can return new signals over time. If the report is still generating there are no signals yet and this returns _status:"processing". This is a free read — it does not consume any credit.

humantic_signals_per_account_v1

ChatGPT
Use this when the user asks what is happening at one specific account - 'any news on Bayer?', 'give me latest signals about datadog', 'how can i get a meeting with CBRE', 'what changed at Wipro this month?', 'give me a reason to reach out to AstraZeneca', 'why now for Carrier?'. Returns recent buying signals for that account, newest first: a headline, what happened and why it matters to the seller, impact (High, Medium, Low), a suggested action, the source link and the event date. Pass the create_request_id of the account's COMPLETED report (find it with humantic_miia_report_list_v1). Filter with date_range (last_7_days, last_1_month, last_3_months, last_6_months; default last_3_months) or since/until, and limit (default 120). This feed keeps only the last 60 days - the response's since field shows the window actually applied; for older history use humantic_signals_all_accounts_v1 with buyer_company. Free - no credit. Returns processing while the report is generating. No report for this account yet? Create one first.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/humantic-ai"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/humantic-ai)