> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sequencehq.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Labs

> Where we prototype features for revenue automation

<Frame>
  <img src="https://mintcdn.com/sequence/_8n3h7QmhHqdBxtd/images/product/labs/labs_cover.png?fit=max&auto=format&n=_8n3h7QmhHqdBxtd&q=85&s=8b5022336fd52323c12cd23a3c650d4d" alt="Sequence Labs" width="1500" height="500" data-path="images/product/labs/labs_cover.png" />
</Frame>

Labs is where we prototype features. These are things we're experimenting with: early prototypes and internal tools, published while we're still building and refining them. Some will never leave the lab; some will shape the future of revenue automation.

<Note>
  Labs projects are released to a small group of customers first while we refine them. If you'd like early access to something here, [speak to us](https://www.sequencehq.com/book-a-demo).
</Note>

## How we build Labs

We see LLMs as useful at three levels, and we build deliberately across all of them:

1. **Slotted into specific product flows**, for example turning a contract into structured data.
2. **Running guided automations with human oversight**: multi-step workflows with tools, review, and audit.
3. **Operating autonomously**, where the guardrails are strong enough to let an agent run on its own.

These principles apply across all three levels, and they matter more the higher you climb:

<CardGroup cols={2}>
  <Card title="The quality of your tools matters" icon="screwdriver-wrench">
    Tools, context, and feedback loops are the biggest differentiators.
  </Card>

  <Card title="Keep humans in the loop" icon="user-check">
    Someone has to see what the agent did, why, and have a path to override it.
  </Card>

  <Card title="Smallest useful scope first" icon="crosshairs">
    Every agent should do one specific job before we widen its remit.
  </Card>

  <Card title="Measure everything" icon="ruler">
    "Vibes are good" isn't enough. Track quality, latency, and cost at every layer.
  </Card>
</CardGroup>

## Prototypes

<CardGroup cols={2}>
  <Card title="MCP server" icon="plug" href="/labs/mcp">
    Expose your Sequence data to AI clients like Claude as a set of tools, so an assistant can fetch and reason over your billing data in response to a prompt.
  </Card>

  <Card title="Contract intake agent" icon="infinity" href="/agents/contract-intake">
    Automatically extract customer details, pricing, and billing terms from a signed contract and set up the customer and billing schedule for review.
  </Card>
</CardGroup>

## Get involved

Labs gets sharper when the people who'd actually use these agents push back.

<CardGroup cols={1}>
  <Card title="Book a demo" icon="calendar" href="https://www.sequencehq.com/book-a-demo">
    See how Sequence agents work with your billing workflows.
  </Card>
</CardGroup>