A non-technical founder's custom platform, built and run by an embedded team
How a founder-led US brokerage with no engineering team got a custom web platform with AI listing tools, built and run by Lyra's embedded engineers for 21 months and counting.
A founder-led US brokerage wanted a custom platform its team could use to present property options to clients, with AI doing the tedious parts. It had no engineers. Lyra became its engineering team: 21 months on, the platform runs the firm's client presentations, AI tools turn a pasted link into a finished listing card, and the work has shipped without the founder managing a single developer.
At a glance
- Client: founder-led US property brokerage, no in-house engineering
- Need: a custom web platform for client presentations, with AI listing tools
- Team: a rotating Lyra team plus a designer over 21 months; one engineer today
- Stack: Next.js, tRPC, Prisma, NextAuth, AWS S3, Redis, Claude API
The problem
The founder knew exactly what the brokers needed but had no technical co-founder and no engineering team. Off-the-shelf tools didn't match how the firm presents options to clients, and a one-off agency build would have left nobody to maintain or extend it.
What Lyra's team built
Boards for client presentations. Multi-board property directories with copy and move between boards, badges, deep links, activity logs and contacts for admins.
AI listing tools. Paste a link and AI creates the listing card. AI also finds duplicate listings, combines cards for single-tenant buildings, and flags buildings that can't be split from the listing's wording.
Data and automation. A scraper for a commercial listings source with map-polygon area search, and a bot that renews classified listings automatically.
Notifications. Email alerts batched every 15 minutes, @mentions, reactions and a first-login tutorial.
Results
- Boards load a fraction of the data they used to. Full-size photos were being sent as thumbnails, so opening a board meant downloading 100 to 200 MB. Cards and map pins now use lightweight versions.
- Email alerts land in the inbox after a provider switch.
- Map costs cut with a Redis cache in front of the maps API.
- Every requested feature shipped, and the work has continued for 21 months.
Software agency vs an embedded engineering team
A software agency builds to a fixed brief and hands it over. That works for a website, but not for a platform a business runs on every day and keeps wanting to improve. This founder needed an engineering team: people who know the codebase, fix problems the same day and add features as the business asks for them. Lyra runs that team, so the founder sets priorities and Lyra handles the engineering.
Read more: What is a forward deployed engineer?
Frequently asked questions
