Skip to content

Details

Abstract:

Everyone has spent the past year wondering whether AI makes their application obsolete. It's a fair thing to worry about, but the worry usually points at the wrong layer. What's hard to replicate isn't the UI. It's the data model, the domain rules, and the workflows your users have built their jobs around.

So here's the question. Your domain model knows things no frontier model does. Can anything reach it? A public API says yes in principle. An MCP server says yes in practice, and if you have the API, you've already built most of one.

Layer is a project management platform where a single project can carry 12,000 interrelated records. New users have to design a schema before the tool does anything for them. Power users know exactly what they want and can't possibly type it all in. We built an MCP server on our existing API and went after both.

What we learned shipping it: the same primitives serve a beginner scaffolding a first project and a power user generating a complex project in an afternoon, and building for an agent exposed API ergonomics problems causing our human users to give up.

Bring your own API. You'll leave with a list of things to go check.

Bio:

Jack is the Lead Software Architect at Layer, where he builds the technical foundation for a platform that connects building data to Building Information Models (BIM). A Nebraska native and graduate of the University of Nebraska-Lincoln’s Raikes School, Jack combines a lifelong passion for solving complex problems with deep technical expertise to help the architecture, engineering, and construction industries work smarter.

Related topics

Events in Lincoln, NE
Artificial Intelligence
Artificial Intelligence Applications
Artificial Intelligence Programming
Computer Programming
Web API Design

You may also like