AI · Overview

AI checklist software that keeps people in charge.

Describe a process and AI drafts the template. Connect the assistant you already use and it can start checklists, fill in fields and complete your tasks. The decisions, and the record of who made them, stay with your team.

Generator free, no signupWorks with any AI assistantNot used to train AI models

In short

Last updated

CheckFlow uses AI in four ways. A free generator turns a sentence into a checklist template, with no account. Inside the app, Generate with AI builds templates from a description or your own documents. The MCP server lets Claude, ChatGPT or any other AI assistant work in your checklists. And REST API v3 with webhooks lets you build your own AI workflows. People review, complete and approve.

Faster to set up, still accountable to run.

A first draft in under a minute

Steps, fields, conditional logic, due dates and halts from a single sentence, ready to edit.

The assistant you already use

Ask in plain language and your assistant does the clicking in CheckFlow, with your permissions and in your name.

People stay accountable

Every change is recorded against a person, and Administrators can switch AI off for the whole team.

Free · No signup

The free AI checklist generator

Type what you need, such as “tenant move-out inspection” or “monthly bookkeeping close”, and watch the template take shape in a preview of the template designer: sections, the fields to record, steps that appear only for certain answers, sign-offs that hold the line and sensible due dates. It takes 15 to 30 seconds.

It checks the template library first, and if a ready-made template fits, you get that instead, written by people. There is no account and no credit card. When you want to edit the template and run it with your team, start a free trial and it is added to your workspace.

  • Free, with an hourly limit to keep it free for everyone
  • Descriptions run on Microsoft Azure and are not used to train AI models
  • Adding a generated template to your workspace uses no AI credits
Open the AI generator

In the app

Generate with AI, inside CheckFlow

In the template designer, Generate with AI builds a template from a description of up to 500 characters, or from up to three of your own documents: Word, Excel, CSV, PDF, web pages, JSON exports from other tools, and photos or scans of paper checklists. Import them as written, or ask it to fill gaps and add good practice too. Inside your workspace it can assign steps to your existing groups.

It is live for every customer and paid for in AI credits shared by the team: 250 a month for each paid seat on Business, 750 on Enterprise and 150 for a free trial. A template from a description typically uses 5 to 10 credits, and one from documents 20 to 40. The result is an ordinary template, saved as version 1, that you check and edit before anyone runs it.

  • Runs on CheckFlow’s own Azure AI resource, in your workspace’s region
  • Documents are read, then discarded, and nothing is used for training
  • Administrators see usage by person and choose which Members can use AI
Generate a template with AI (docs)

MCP server

Your AI assistant, working in CheckFlow

The CheckFlow MCP server connects Claude, ChatGPT, Gemini, Grok, Copilot or any other AI assistant to your workspace. Ask “what’s on my plate this week?”, “start an invoice review for INV-2041 and assign the approval to Sarah” or “add these twelve suppliers to the Suppliers Data Set”, and it does the work through 147 tools covering checklists, tasks, standalone tasks, templates, Data Sets, schedules and webhooks.

It connects with a CheckFlow API key and acts as the member the key belongs to: it sees what they can see, and its changes are recorded in their name. Assistants run on their own provider’s AI, so they use no AI credits. The docs give step-by-step setup for Claude Code, Claude Desktop, Cursor and VS Code, and any client that can send a custom header can connect.

  • Tools are marked read-only, writing or destructive, so your client can ask before deleting
  • It can build and refine templates over a conversation, including notifications, Data Sets and linked checklists
  • Included on every plan
Read the MCP guide
Connect your assistanthttps://api.checkflow.io/mcp

Claude Code

claude mcp add --transport http checkflow https://api.checkflow.io/mcp \
  --header "X-API-KEY: your-api-key-here"

Cursor · ~/.cursor/mcp.json

{
  "mcpServers": {
    "checkflow": {
      "url": "https://api.checkflow.io/mcp",
      "headers": { "X-API-KEY": "your-api-key-here" }
    }
  }
}
Use a key that acts as you, and keep it in your client’s secret settings.

For developers

Build your own AI workflows on the API

The MCP server and REST API v3 are the same operations. Use the API when you are writing the agent or the integration yourself.

  • REST API v3

    Start checklists, complete tasks, answer fields, build templates from a JSON document and keep Data Sets in sync, with idempotent writes for safe retries.

    Developer overview
  • Webhooks

    Nine signed events, from a completed task to a changed Data Set record, so your agent reacts when something happens instead of polling.

    Webhook events (docs)
  • A guide written for models

    The template authoring guide, at GET /v3/schema/guide and as an MCP resource, tells a model how to compose a template that validates.

    Template documents (docs)

Which to use

Generator, Generate with AI or the MCP server?

They overlap on purpose. Pick by what you have to start from and how much back-and-forth the process needs.

Choosing between CheckFlow’s AI options
QuestionFree generatorGenerate with AIMCP server
Account neededNoYesYes, and an API key
Starts fromA sentenceA sentence or up to three documentsA conversation
Knows your teamNoYour groupsYour members, groups, tags and templates
Can run checklists and tasksNoNoYes
Paid inFreeAI creditsYour assistant’s own plan
Best forA quick first draftTurning existing SOPs into templatesComplex processes and day-to-day work

Generate with AI, the MCP server and the API are included in the Business plan at $10 per user per month, or $9 billed annually.

Honestly

What AI does, and doesn’t do, in CheckFlow

  1. 01

    AI drafts templates. People check them.

    Every generated template is an ordinary draft you read and edit before anyone runs it. AI makes mistakes, so compare an imported template with the document it came from.

  2. 02

    AI is never an assignee.

    Tasks are assigned to people and groups. An assistant connected over MCP acts as the member whose API key it holds, and everything it does is recorded in that person’s name.

  3. 03

    It cannot reach further than the person.

    An assistant is refused wherever its member would be: a step assigned only to someone else, a task behind an incomplete halt, a template they cannot edit.

  4. 04

    Administrators decide whether it is on.

    Allow AI features and Allow AI assistants to connect are separate switches on the AI tab. Members need the AI.User permission, and Guests can never use AI features.

  5. 05

    Your data is not used to train models.

    Generate with AI sends only what it needs, runs on CheckFlow’s Azure resource in your region, and keeps a record of who used AI and how much, never of what was asked.

  6. 06

    Analytics is not an AI feature.

    Cycle times and bottlenecks come from Analytics in the app. Neither the MCP server nor API v3 has an analytics tool.

What an assistant reads through the MCP server is handled by that assistant’s provider, under your organisation’s agreement with them.

MCP in practice

Ask, and it happens in CheckFlow.

An example request, the real tool the assistant calls, and what it can truthfully say back.

  • Start with a draft

    Describe the process and get a template in under a minute, free and without an account.

    Try the AI generator
  • Connect your assistant

    Setup takes a few minutes in Claude Code, Claude Desktop, Cursor and VS Code, and any client that sends a custom header works.

    Connecting a client (docs)
  • Or build your own

    REST API v3 and webhooks give your own agent the same operations.

    Developer overview
AI assistantcheckflow MCP · connectedIllustrative example

Start an IT offboarding for Alex M. Their last day is Friday 23 October.

checkflow · create_checklistdone
template: "IT Offboarding SOP"
name:     "IT Offboarding – Alex M."
Last day: "2026-10-23"
→ 201 · 15 tasks · assigned from the template

Done. IT Offboarding – Alex M. is running with 15 tasks, assigned to IT, HR and Alex’s manager, with due dates counted from Friday 23 October.

AI questions

What can AI do in CheckFlow?

Draft templates and work alongside you. The free AI generator and in-app Generate with AI build checklist templates from a description or your documents. Through the MCP server, an AI assistant can start checklists, fill in fields, complete tasks, build templates and update Data Sets for you. Completing and approving work stays with your team’s members.

Which AI assistants work with CheckFlow?

Any AI assistant that can connect to an MCP server, including Claude, ChatGPT, Gemini, Grok and Copilot. The docs give step-by-step setup for Claude Code, Claude Desktop, Cursor and VS Code. A client that can only authenticate with OAuth can connect through the mcp-remote bridge.

Is the AI generator really free?

Yes. The public generator needs no account or credit card, with an hourly limit to keep it free for everyone. Inside CheckFlow, Generate with AI uses the AI credits included with your plan.

How do AI credits work?

Generate with AI is paid in credits shared by the team: 250 a month for each paid seat on Business, 750 on Enterprise, and 150 for a whole free trial. A template from a description typically uses 5 to 10 credits, and one from documents 20 to 40. If the team runs out, an Administrator can buy a pack, from 2,500 credits for $10, and bought credits last 12 months. The MCP server and the REST API use no credits.

Is our data used to train AI models?

No. Generate with AI runs on CheckFlow’s own Microsoft Azure AI resource in your workspace’s region, using an OpenAI model hosted by Microsoft. Microsoft does not use prompts or results for training, nothing is sent to OpenAI, and documents are discarded after reading. What an MCP assistant reads is handled by its own provider.

Can an AI assistant approve or complete work on its own?

An assistant acts as the member whose API key it holds, so it can do what that person could, and every change is recorded in their name. It is refused where they would be, such as a step only someone else can complete. Use a key that acts as the least powerful member who can do the job, and keep your client’s confirmation prompts on.

Can we turn AI off?

Yes. On the AI tab of Team Management, Administrators can turn off AI features and AI assistant connections separately, for the whole team, and choose which Members hold the AI.User permission.

Which plans include the MCP server?

The MCP server, the REST API and webhooks are included on every plan, starting with Business at $10 per user per month, or $9 billed annually, and in the free trial. See pricing.

Let AI do the drafting. Keep your team in charge.

Try the free generator now, or start a trial and connect your AI assistant today.

14-day free trialNo credit card required