How a developer-tools SaaS built a dedicated development team with Lyra
How an established developer-documentation SaaS used a dedicated development team from Lyra to ship AI features, MCP support, security hardening and a new editor, growing from one embedded engineer to about fifteen.
An established, venture-backed developer-documentation platform had one engineer covering AI features that needed two or three. Lyra embedded a dedicated development team that grew to about fifteen engineers in sixteen months, shipping AI products, MCP support, security hardening and a rebuilt editor, with its own Lyra lead managing the pod inside the client.
At a glance
- Client: established developer-documentation SaaS, San Francisco
- Need: AI product capacity, then platform, security and editor work
- Team: first engineers June 2025, about 10 by April 2026, about 15 by mid-2026
- Stack: TypeScript, React, Node, LLM APIs, Langfuse, Cloudflare Workers, MCP
The problem
The company's AI team was under-resourced: one engineer was covering features that needed two or three. Meanwhile the core product still needed security, billing and editor work. Hiring a dozen senior engineers in San Francisco would have taken a year.
What Lyra's engineers built
AI features. AI translation that patches only the changed parts of a page instead of re-translating it, an AI content importer, an AI-readiness checker for documentation, a unified AI search, and an AI writer for Slack with its own evaluation suite.
MCP and agents. OAuth sign-in for the company's MCP server, usage metrics for MCP and agent traffic on a Cloudflare Worker, and internal MCP tools for the sales and support teams.
Security hardening. A central permission checker, read-only API keys and two-factor step-up for sensitive actions.
Editor and collaboration. WYSIWYG and collaborative editing, review comments that can block a merge, and Slack and Teams notifications.
Results
- AI importer shipped to general availability, and the Slack AI writer launched in October 2026.
- Translation cost and latency down sharply with the patch-based approach.
- A slow endpoint cut from 5 to 7 seconds to about 1 second.
- Editor phase one rolled out across the product.
- One engineer to about fifteen, led by a Lyra engineer who manages the pod day to day.
Why a dedicated development team instead of a software development agency
An agency takes a brief and returns a deliverable. A dedicated development team from Lyra joins your Slack, your repos and your standups, picks up your conventions and stays. That is what let this team move from AI features to security and editor work without a new statement of work each time, and why the client kept adding engineers.
Read more: What is a forward deployed engineer?
Frequently asked questions
