Skip to main content
PickFu gives you four ways to create surveys, pull results, and build consumer feedback into your workflow — from clicking through the web app to calling the REST API. The MCP server is the fastest way to get started in code: it works with the AI assistants you already use. However you work — the web app, an AI assistant via the MCP server, the CLI from scripts and agents, or the REST API from your own app — you reach the same PickFu panel of real people, who return votes and written explanations in minutes. However you work — the web app, an AI assistant via the MCP server, the CLI from scripts and agents, or the REST API from your own app — you reach the same PickFu panel of real people, who return votes and written explanations in minutes.

​
Which should I use?

Read the Reach for it when row first — it maps your situation to a tool. Everything below it is supporting detail.
Web appMCP serverCLIREST API
Who it’s forAnyone — no codePeople working inside an AI assistantDevelopers, scripts, autonomous agentsDevelopers building integrations
Reach for it when…You want to create or read a poll by hand, visuallyYou want to run research by chatting with Claude, Cursor, or ChatGPTYou need polls in scripts, CI/CD, or a watch loop for agentsYou’re embedding PickFu in your own app, backend, or a no-code tool (Zapier, Make)
How you use itOpen app.pickfu.comAsk your AI assistantpickfu survey create … in a terminalHTTP requests to api.pickfu.com/v1
Getting startedJust sign in — nothing to installConnect once (~1 min) · OAuth or API keynpm i -g @pickfu/cli · OAuth or API keyCreate an API key
OutputVisual dashboardRich chat — text + inline imagesJSON + readable tablesJSON (OpenAPI-spec’d)
Start with the web app if you just want results. Add the MCP server to drive PickFu from the AI assistant you already use. Reach for the CLI when you need automation or agents. Call the REST API when PickFu lives inside your own product.

MCP server

Connect Claude, Cursor, ChatGPT, Notion, Windsurf, and other AI assistants to PickFu.

CLI

Create and manage polls from the command line.

REST API

Call PickFu’s v1 endpoints directly from any language.

Authentication

OAuth and API key authentication for the MCP server, CLI, and API.

Support

Get help from our support team.

​
What you can do

With the PickFu MCP server, CLI, and REST API, you can:
  • Create surveys from AI assistants — Ask Claude, Cursor, or ChatGPT to create and publish PickFu polls through natural conversation
  • Generate and upload media — Create images from prompts or upload your own for use in polls
  • Retrieve real-time results — Get survey responses, AI summaries, and analytics as they come in
  • Target specific audiences — Use demographic traits to reach the right respondents
  • Automate your workflow — Integrate PickFu into your product development pipeline

​
Still choosing between MCP and CLI?

These two are the closest substitutes — both let AI tools drive PickFu over the same platform. If you’ve narrowed it down to these two, here’s where they actually differ:
CapabilityMCP serverCLI
View survey images inline (in AI assistant chat)Yes–
Poll URL builderYes–
Help center searchYes–
Watch/wait for survey completion (survey watch)–Yes
Upload media from a local file–Yes
JSON output for scripting–Yes
Schema introspection–Yes
Everything else is shared: both create, update, publish, stop, and delete surveys; add respondents; read responses with demographics; generate images and upload media by URL; and manage tags, projects, and playbooks.
Use the MCP server for conversational AI assistants like Claude, Cursor, and ChatGPT.Use the CLI for scripts, CI/CD pipelines, when Claude Code needs survey watch for blocking loops, or when agentic AI tools with terminal access (OpenClaw, Devin) need local file uploads or lower token usage. The CLI is more efficient for fully autonomous workflows.

​
Getting started

1

Connect the MCP server

Add PickFu to your AI client — setup takes about a minute. The MCP server works with Claude, Cursor, ChatGPT, Notion, Windsurf, and any MCP-compatible client.
2

Authenticate

Choose your auth method: OAuth (automatic — sign in when prompted) or API key (create at Settings > API Keys for automation and CI/CD). Both work with the MCP server and CLI.
3

Create your first poll

Ask your AI assistant to create a PickFu poll, or use the CLI to launch one from your terminal.