Gusto Embedded recently launched our Developer MCP Server in open beta. This MCP (Model Context Protocol) server lets a developer’s AI assistant—like Claude, Cursor, or Copilot—pull live, version-aware Gusto API docs into the IDE where they’re already working. We sat down with Ian Gawronski, a member of our Technical Solutions team, to talk about what it does, how it’s already changing the way partners build, and where it’s headed.
What is the Gusto Embedded Developer Assistant MCP Server, and what was the idea behind it?
“It allows developers building (or considering) Gusto embedded payroll to find answers around API questions or best practices without having to read through massive chunks of our documentation.”
The idea, Ian explains, is to serve different kinds of developers exactly what they need. Frontend engineers can use it to understand the shape of API responses and mock them out before they even connect the backend. But the deeper value shows up for backend developers.
“Let’s look at state taxes. There is a lot of complexity with state taxes. The Developer MCP could help you understand the structure of the state taxes API response so that your proxy service and SDK are handling issues like multi-jurisdiction edge cases correctly. It allows backend developers to quickly get API context in a way that’s not just reading text. It’s actually giving you examples and tailored code snippets.”
How has the MCP server changed the way our partners interact with our documentation day-to-day?
The team behind the Developer Assistant MCP didn’t just hand it to partners–they’re also some of its biggest users.
One example stuck out. A solutions team member was trying to debug a partner’s issue in Postman: a race condition where two requests fired almost simultaneously. Recreating it manually was painstaking. With the MCP-powered set of test scenarios, she could reproduce it immediately.
“She was able to use the MCP to debug and identify the edge case causing issues for the partner, where before she would have had to go to Engineering and do lots of trial and error to figure out what was going on.”
When a new teammate joins one of our partners’ payroll teams, how should they use the MCP?
Ian joined Gusto Embedded shortly before the MCP launched and used it throughout his own ramp-up. He came to Gusto with nearly a decade of engineering experience, which meant he was learning payroll from the inside out, through the API and back-end functionality rather than starting with payroll domain knowledge. The MCP let him fill gaps without constantly pulling colleagues away from their work.
“I ramped up faster than I expected to—I was able to ask the MCP what I needed instead of pinging a teammate every time I hit something unfamiliar.”
“When a coworker went out on leave, I was tasked with helping backfill on a key account. I had to get up to speed quickly, and that was one way to do it without bugging other team members all the time.”
The same applies for partners onboarding new developers to an existing payroll integration: the MCP can help new engineers get up to speed on API and payroll system architecture best practices, and understand what responses look like without having to read through all of the documentation.
You’re currently working closely with a partner on a significant build. How are they using it?
Ian has been embedded (no pun intended) with an enterprise partner navigating an aggressive build timeline to be ready for the payroll switching season spike at the end of the year. Their backend developer has already been leaning on the Developer MCP, and Ian’s approach is to keep reinforcing that habit.
“Right now, I’m taking questions, plugging them into the Developer MCP, and making sure it can answer those questions effectively. Over time, I’ll be working with our partners’ developers to make sure that the MCP is the first place they ask questions as they build.'”
The Dev Assistant MCP differs from adding a link in your AI coding tool to the existing Gusto Embedded docs by applying schema awareness, limiting the need to copy-paste snippets of code, and running right inside the tool you’re already using to build payroll.
The partner is building with Gusto Embedded’s SDK, which Ian thinks makes the MCP more useful.
“They may end up building and launching payroll even faster because they’re not trying to build everything custom using answers from the Developer MCP. They’re leaning on the SDK, which handles a significant portion of edge cases automatically, and using the MCP to fill in the gaps”
Where does the MCP server excel, and where are its limits?
Ian is candid about what the MCP does well and where it hits walls.
The MCP is not a substitute for payroll expertise. Because it’s grounded in documentation, it reflects the documentation’s strengths, but isn’t an end-all solution for building, selling, and supporting payroll (at least not yet).
“You still have to be someone who has a decent amount of payroll knowledge. It’s not like a set it and forget it thing. You want to make sure you’re reviewing the outputs. Like any AI tool, particularly in a domain as consequential as payroll, it’s something you need to stay engaged with.”
That said, for the bulk of what partners encounter during an active build, it holds up. And when partners run into any issues, they have Gusto’s technical team ready to help them. The developer assistant MCP server doesn’t replace human review, where partners can lean on a team that’s been building payroll infrastructure for 14+ years.
Is the Developer Assistant MCP server useful for existing partners as well?
Absolutely, Ian says. The documentation it draws from stays current, which matters more than developers might expect.
“Our documentation is changing all the time. We have a lot of releases, and for major releases we announce the most important updates. But as we fix bugs or issues, there may be changes in documentation that a partner may not be familiar with because they’re used to working on a version from a year ago. The MCP makes it easier to stay up to date on every change reflected in our documentation, so you’re always getting the most relevant information.”
What’s next for embedded payroll API docs?
As AI tools grow in sophistication, so too will future iterations of our Dev Assistant MCP server and other AI-enabled developer functionality. While it’s too early to share any specifics on that front, the Dev Assistant MCP server will make it easier than ever build and launch payroll. From there, partners can lean on Gusto’s 14+ years of experience help with the rest: tax and compliance, money movement, and best-in-class enablement to help our partners market, sell and support their payroll customers.
Try it yourself
The Gusto Embedded Developer Assistant MCP Server is available in an open beta. If you’re building on our platform or exploring what it would look like to add payroll to your product from a technical perspective, that’s a great place to start, alongside our existing developer docs.
To start using the Developer Assistant MCP server, use the JSON snippet below in Claude code. For other AI coding tools, refer to our documentation.
{
"mcpServers": {
"embedded-payroll": {
"url": "https://embedded-payroll.readme.io/mcp"
}
}
}
For more on how Gusto Embedded is built for developers, explore our other developer blog posts.
