Homebase Scaled Governed AI Workflows to 400+ Employees in Under 5 Months With Runlayer

Homebase is the AI-powered platform for hourly teams, with employee scheduling, time clocks, payroll, hiring, communication, HR, and more. More than 150,000 small businesses rely on Homebase to make work radically easy and give their teams superpowers. As the leader in small business team management, Homebase tracked 1+ billion hours for 3.5+ million workers last year. Homebase was named a 2024 Fast Compa’s Brands That Matter, a 2025 Webby Award winner, and a 2025 Fast Company’s Most Innovative Workplaces North America winner.

As Senior Manager of AI Acceleration at Homebase, Justin Nazari is constantly seeking opportunities to leverage AI as a force multiplier across the company’s expanding workflows. However, as Homebase expanded its services, low control & visibility into MCP, agent, and skill usage across the company threatened its path to scalable AI transformation.

Justin knew that allowing AI unchecked access to employee information, payroll, and finance data without control & visibility was a non-starter. Yet several engineers and power users had already set up local MCPs to streamline adjacent workflows, connecting them to tools such as GitHub, Datadog, and Slack without complete oversight. While these early adopters demonstrated a strong appetite for more AI-assisted workflows, Justin couldn’t allow innovation to bypass security and control.

Unsanctioned experimentation was far from the only blocker to scalable AI transformation at Homebase. At the time, there was no standardized path for Homebase employees to request new connectors, access internal data sources, or reuse workflows teams had already built. As uncertainty around responsible use grew, non-technical teams (such as marketing and operations) had fewer practical ways to leverage AI in their day-to-day work. "Folks at Homebase wanted to benefit from what AI-enabled workflows could offer, but we needed to look to security as the enabler, not the roadblock,” Justin explains.

To regain visibility and control over AI experimentation at Homebase, Justin considered building an in-house MCP gateway. However, building a gateway that met strict corporate security and compliance standards would require creating granular RBACs and in-depth audit logging, steps that would tie up engineering resources for months and require continuous, active maintenance.

Rather than reinventing the wheel, Justin partnered with Runlayer. The platform offered not just an MCP gateway, but also a full control plane for AI, a slack-native agent builder, a skills & plugins registry, and much more.

The partnership began with a quick and collaborative onboarding. Justin worked directly with the Runlayer team to establish SSO via Okta, then launched an approved registry of Runlayer-governed connectors across core systems such as Datadog, Databricks, CrowdStrike, Sentry, GitHub, Slack, Figma, and Google Workspace. “Runlayer’s engineers were in our Slack comms daily,” Justin shares. “Right out of the gate, they felt like an extension of our team.”

From there, Justin and the broader team began deploying custom company-built MCPs through Runlayer, with access scoped by user, team, task, initiative, and prototype. Every MCP, skill, and agent is now housed within a unified registry that’s accessible to employees across the company. On the backend, Runlayer automatically monitors for misuse or prompt injection and maintains centralized audit logs so security and enablement teams can understand how each connector or agent is being used, with a full agent lifecycle trace.

With Runlayer, Homebase employees now have secure access to AI that actually works and fits their unique workflows. Engineers can pull Slack threads, check Linear tickets, query Datadog, read Confluence, and write a postmortem all without leaving Claude Code. Business teams use Claude against approved connectors without self-configuring local MCPs or managing separate credentials. And security, IT, and enablement teams can support broader AI usage, knowing that approved connectors, MCPs, skills, plugins, and agents are running through a platform built for centralized control.

Crucially, Runlayer unlocked new paths for non-technical teams to securely adopt AI-assisted workflows. Not only can these teams submit new connector requests via Slack in seconds, they can also describe an agent in natural language, then have it built with the appropriate permissions, identity, and security characteristics applied automatically. Now, even the most non-technical employees across the company can build agents in one sentence, governed automatically. In one instance, a marketer who spent months building automations in his free time evolved into a dedicated “agent builder” for his team.

With Runlayer, Homebase turned an environment of disparate AI experimentation into a collaborative, managed ecosystem that leverages safety as an accelerator rather than a roadblock. Today, employees across the company safely connect their AI tools to approved data sources and services, deploy and share MCP servers, skills, and agents, and automate tedious cross-system workflows with the confidence that everything is governed and auditable.

Looking ahead, Justin is eager to continue evolving the partnership from a foundation of widely accessible MCP servers into a full‑fledged AI operating model built on skills and agents.