Skip to content
MCP Chats

For teams putting AI agents to work

Your business processes, run by AI agents.

Describe how the work should go. Your AI agent follows it: it uses your data and systems, sends your emails and asks the right person before anything important. Every step is on record.

Join the waitlistSee how it works

No code, and no new AI to buy. Works with Claude Desktop, Codex, Cursor, OpenClaw and 15 more. Free during the pilot.

Run by your AI agent

Refund request · order #5521

Your agent is working

  1. Your AI agent

    Read the order

    Looking up order #5521…

  2. Jev · decision AI

    Assess the claim

    Next

  3. Your AI agent

    Apply your refund policy

    Next

  4. Priya · Finance

    Approve the refund

    Next

  5. Your AI agent

    Issue the refund

    Next

  6. Your AI agent

    Tell the customer

    Next

On the recordData readAI judgementRule matchedApprovalSystem updatedEmail sent
  • Your agent
  • MCP Chats
  • A person
  • Done
  • · Illustrative

AI agents can do the work. Trusting them with it is the hard part.

Most teams stop at the same questions. Watch how each one is answered inside the process.
  • Without a processIllustration

    Done! Refund issued. ✓

    Issue the refund · required to finish

    • Refund receiptMissing
    • Amount matches the orderNot checked

    Sent back to the agent

From “how we do it” to done, in three steps.

You don’t write code or switch AI tools. You decide how the work should go, and what your agent may use along the way.
  1. Step 1

    Describe how the work goes

    Tell your agent, or draw it on the canvas: the steps, the rules, the judgement calls, who signs off and what counts as done.

    Start · refund request
    Your AI agentRead the order
    Jev · decision AIAssess the claim
    FinanceApprove the refund
  2. Step 2

    Connect what it needs

    Your databases, your business systems and your email templates. You choose what the agent may read and change.

    PostgreSQLSQL ServerMongoDBCosmos DBAny REST APIYour email templatesYour rules
  3. Step 3

    Your agent does the work

    In the AI agent your team already uses. People sign off by email, and every step lands on the record.

    Your AI agentOrder read · $740.00

    Your AI agentOver $500 · to finance

    Priya · FinanceApproved, from email

    Your AI agentRefund issued · email sent

Judgement calls, made the same way every time.

Some steps need judgement: what a customer is asking for, whether a draft meets your standard, whether an application is complete. Jev, the decision AI from TypeSafe built into MCP Chats, answers the question you set, against your criteria.
Sort
“Which kind of request is this?” Picks one of your categories, so the process takes the right path.
Score
“How well does this meet our standard?” Rates work on your own scale, from your own descriptions.
Check
“Is this true, given the evidence?” Says how likely a statement is, so weak cases don’t slip through.
  • When Jev is unsure, a person decides. Unclear never passes on its own.
  • You set the lines. Your rules decide what each answer means.
  • Agents can’t change its answers, and each one is kept with the run.

Jev · decision AI

Your question: what is this request about?

Refund request 1 of 4

“The vase arrived in pieces and the box was crushed.”

  • Damaged in transit96%
  • Changed their mind3%
  • Never arrived1%
  • Your line: 70% sure

Damaged in transit

Refund path

0

Changed their mind

Returns policy

0

Unsure

A person decides

0

Illustrative example. You set the question and the line.

Everything the work needs, in one place.

Each part works on its own, and they work best together.
Your processes·Your data and systems·Your emails

No new AI to buy.

Keep the AI agents your team already uses. Connect them once, and your processes, data and emails are ready in each one.

Agents connect through MCP, the open standard AI agents use to plug into business tools. Setup takes a few minutes, and we help you during the pilot.

See all 19 supported agents

Know exactly what happened, and who decided.

Open any run to see what the agent did, the data it used, how Jev judged it, the rule that decided, who approved and which emails went out.
  • Passwords and keys never reach the agent.
  • Each organisation’s data is kept apart.
  • Every approval keeps what the person saw.
Security and boundaries

Illustrative run record

Refund request · order #5521

Completed

  1. Your AI agent · orders database

    Order read

    Order #5521 · $740.00 · arrived damaged

  2. Jev · decision AI

    Claim assessed

    Damaged in transit · 94% sure · above your 70% line

  3. Your AI agent

    Route chosen

    Your policy says over $500 goes to finance; reason recorded

  4. Priya · Finance

    Approved, from email

    Decision + what Priya saw, kept together

  5. Your AI agent · payments system

    Refund issued

    Receipt re_8f2

  6. Your AI agent · your template

    Customer told

    “Your refund is on its way” · sent once

What the pilot supports today
  • Agents connect through MCP with an access token. ChatGPT and claude.ai connectors and scheduled starts are not enabled in the pilot.
  • Controls apply to work routed through MCP Chats. We do not control an agent’s other tools or guarantee the truth of its reports.
  • Database sources support reads. Enabled REST endpoints support writes. Published email templates support outgoing communication; handling replies is not supported yet.

Questions people ask.

Short answers. The product pages go deeper.

What is MCP Chats?

MCP Chats turns your business processes into steps the AI agents you already use can follow. Your agent does the work with your data, systems and email templates, Jev makes the judgement calls against your criteria, and the right person signs off before anything important. Every step is recorded.

Which AI agents does it work with?

19 agents, including Claude Desktop, Claude Code, Codex, Cursor, OpenClaw, VS Code and Gemini CLI, each with a step-by-step setup. They connect through MCP. Connecting Claude and ChatGPT by signing in is coming with the hosted pilot.

How is this different from approving actions in my agent's chat?

Your agent's chat asks you to allow one action at a time, and the decision scrolls away with the chat. MCP Chats asks the right person to approve the result, with the evidence attached, from an email link if they like. The run and the decision stay on record, and the same process works in any supported agent.

Do I need to write code?

No. Describe the process to your agent or draw it on the canvas, connect a database or import an API definition, write your email templates and choose who signs off. During the pilot, we build your first process with you.

Can the AI agent see our passwords or change our data freely?

No. Passwords and keys stay encrypted with MCP Chats and never reach the agent. Databases are read-only, and an API action that changes data works only after an admin enables it. These limits cover work routed through MCP Chats; tools your agent has elsewhere are outside our control.

What is Jev?

Jev is the decision AI from TypeSafe, built into MCP Chats. It makes judgement calls inside a process: sorting a request into your categories, scoring work on your scale, or checking whether a statement holds. When it's unsure, a person decides, and agents can't change its answers.

What is MCP?

The Model Context Protocol (MCP) is the open standard AI agents use to connect to tools and data. MCP Chats gives your agents one MCP server with your processes, data, email templates and approvals.

How much does it cost?

It's free during the pilot. Join the waitlist and we'll invite you in small groups and help you build your first process. We agree any pricing with you before a paid plan starts.

A free pilot, with help getting started

Bring one process. We’ll build it with you.

Tell us about a job you’d like an AI agent to handle. When you’re invited to the pilot, we’ll set it up together and run it with your agent.

Join the waitlist

For solo operators and teams. Free during the pilot.

  1. Step 1

    Pick one recurring job

    Refunds, supplier onboarding, client reports or content reviews: work someone does by hand today.

  2. Step 2

    Build the process together

    On a setup call, map the steps, connect your data and emails, and choose who signs off.

  3. Step 3

    Run it and shape what comes next

    Connect your agent, review the results and tell us where the process needs to improve.