Feedback collection, AI-native

Use casesFAQ

For community managers, brand teams, indie makers

Ask your audience anything. Let your AI run the loop.

Stop building forms. Tell Claude (or any agent) what you want to learn from your members, customers, or event attendees — it designs the survey, hands you a share link, and reports back once responses land.

Add to Claude CodeSee use casesDocs
Claude Desktop creating a survey with HumanSurvey MCP, and the resulting form on mobile

Get started

claude mcp add humansurvey -- npx -y humansurvey-mcp

No API key needed — the agent creates one on first run.

The loop

No more Typeform. No more chasing responses. No more pivot tables.

Here's what running a community feedback loop looks like once your AI can do it for you:

  1. Friday

    You

    You just wrapped an AMA with 400 community members. Great session — now you want to know what they thought.

  2. Monday

    You → Claude

    "Send everyone who attended a 3-question feedback form — rate the session 1–5, one thing to change, the topic they want next."

    Claude writes the schema, creates the survey, hands back /s/abc123. You drop the link in #general. Or ask Claude to post it for you.

  3. Tuesday

    You → Claude

    "What did people think?"

    "184 responses. 4.3/5 average. 60% want an infra-scaling deep dive next. Top complaint: too many interruptions during live Q&A."

HumanSurvey returns you a share link. You (or your agent, if it has a Slack/Discord/email tool) post it where your audience already lives — we don't email-blast for you. That's by design.

Who uses it

Whoever needs to hear from a crowd outside their company.

The unifying shape: an audience of members, customers, or attendees — answers that can arrive over hours or days — a synthesis your AI can act on.

Community / brand manager

After an AMA, campaign, or product drop

  1. 01Tells AI: "ask attendees to rate the session + what they want next"
  2. 02Gets a link, drops it in Discord / Slack / email
  3. 03Next day, AI reads back ratings, themes, and top complaints
// schema sketch
title: "Community AMA feedback"
scale · Rate the session (1–5)
multi_choice · Topics you want next?
text · One thing to change next time?

Indie maker / PM

A week after a new-product launch

  1. 01Tells AI: "survey our first 200 users — why they signed up, top paper cut"
  2. 02Agent creates the survey; link goes in the welcome email
  3. 03AI returns ranked pain points; roadmap issues auto-updated
// schema sketch
title: "What should we ship next?"
multi_choice · Top 3 features you’d use
text · Biggest paper cut so far?
scale · Likelihood to recommend (0–10)

Event organizer

After a conference, meetup, or webinar

  1. 01Tells AI: "rate each session, capture speaker feedback, collect next-event suggestions"
  2. 02Link sent via the event app or post-event email
  3. 03AI writes the retro with session-level breakdowns
// schema sketch
title: "Post-Event Feedback"
matrix · Rate each session
scale · Overall event rating (1–5)
text · What should we do differently?

When this fits

Reach for it when

  • You want feedback from your community, customers, attendees, or waitlist
  • Responses can arrive over hours or days — async is fine
  • The questions are known up front: ratings, choices, short text
  • You want your AI to consume the results and do the next thing

Use something else when

  • 1-on-1 clarification with a single user (just chat)
  • Long-form interviews or open-ended research transcripts
  • Analytics dashboards for a human PM to browse
  • Lead-gen forms meant for a marketing automation funnel