Use OAS Files and Your AI Coding Tool of Choice to Build on Empower

The 101: What does this mean for you?

We shipped something this week that’s going to make integrating with Dark Matter a lot easier. Every API on our Developer Platform now has a downloadable specification file. One click, full documentation, machine-readable.

If you’re not a developer, here’s what that means in practical terms. Software systems talk to each other through APIs. Building those connections used to require a lot of back-and-forth between teams. Your developers are asking ours how something works, our team writing custom documentation, your team interpreting it, both sides fixing the gaps. Weeks of project time.

A downloadable spec file replaces most of that. Your developers download one file, and it tells their tools everything they need to know about how to connect to our platform. They can generate the code, test the connection, and start building. No phone calls. No documentation hunts. No waiting for our team to interpret what something does.

For lenders, this means faster integrations and lower implementation costs. For developers, it means we treat your time the way you’d want us to. For the industry, it’s one more signal that mortgage technology is finally catching up to how modern software gets built.

Now the details: what’s actually in the box

The file format we’re publishing is called an OpenAPI Specification, or OAS. It’s the industry standard for documenting REST APIs, used by Stripe, Twilio, Plaid, and most of the companies’ developers love to build with. We’re publishing OAS 3.0 files for every API on the Developer Platform.

Here’s what’s inside one of those files. Every endpoint we expose, meaning every URL your code would call to do something on our platform, is documented in machine-readable form. The request format, the response format, the authentication requirements, the error codes, the data types, the field-level validation rules. All of it.

That sounds like a lot, and it is. But the reason developers care is what they can do with that file once they have it.

Auto-generate SDKs. Tools like OpenAPI Generator can read the spec file and produce a software development kit in whatever language your team uses. Python, Java, TypeScript, Go, C#, Swift. Instead of your developers writing the connection code from scratch, they generate it in minutes. That code is consistent, well-tested, and stays in sync with our actual API.

Import into testing tools. Postman, Insomnia, Bruno, and similar tools accept OAS files as direct imports. Your developers can stand up a full testing environment against our API in the time it takes to drag a file into a window. That used to be a multi-day setup project.

Stand up mock servers. A spec file lets your developers build against a fake version of our API that behaves exactly like the real one. They can write code, run tests, and verify their integration before they ever hit our production environment. That saves real money on cloud costs and real time on debugging.

Validate documentation against reality. This is the one your developers will appreciate most. Some platforms publish documentation that’s been written by hand and slowly drifts out of date as the API evolves. OAS files are generated from the actual API code. The spec is the contract. If the spec says an endpoint behaves a certain way, that’s how the endpoint behaves.

Building on Empower with your choice of AI coding tools

OpenAPI specs work especially well as input for AI coding assistants. Drop our spec into Claude, Cursor, ChatGPT, or GitHub Copilot, and tell it what you want to build. The AI reads the spec, understands every endpoint, knows the data shapes, and writes working integration code against our platform in minutes. What’s great is, this is not pseudocode. This is functioning code that compiles, runs, and talks to our API correctly the first time.

What used to be a multi-week engineering project for a lender’s IT team is now a conversation. “Read this spec. Build me an endpoint that pulls every loan in underwriting status and posts a notification to our Slack channel.” The AI writes it, then your developer reviews it, and you ship it.

The OAS file is the bridge that lets AI coding tools extend Empower as fast as your team can describe what they want.

We’re publishing the spec partly because we know this is how developers are working in 2026. The teams building the fastest aren’t writing integration code from scratch. They’re directing AI to write it for them against open standards, and we’re giving them the standard.

For lenders, this means the gap between “we wish Empower did X” and “Empower does X” can compress dramatically. Custom workflows, internal tools, partner integrations, reporting pipelines. Anything you can describe to an AI coding assistant, your team can now build on top of our platform without waiting on us.

That’s an AI-extensible platform. Not just AI features inside the product. AI extensibility on the platform itself.

What this unlocks for lenders building with us

If you’re a lender integrating Dark Matter into your existing technology stack, this changes three things about how the project goes.

The first is timeline. Integration projects that used to take months can compress. Your developers download the spec, generate the client code, and start wiring things up against real endpoints from day one. The discovery phase that used to dominate the front of these projects shortens significantly.

The second is cost. Less back-and-forth with our integration team means lower professional services hours on your side and ours. The spec is doing the work that humans used to do.

The third is independence. You don’t need our team to interpret what our platform does. The spec is the answer. Your developers read it directly; the same way they’d read the documentation for any modern SaaS API. We become a platform you build at your pace, not a vendor whose timeline you wait on.

What this unlocks for partners and developers

If you’re a technology partner building in title, closing, appraisal, credit, AI tooling, or anything that connects into the mortgage workflow, the OAS file changes the calculus on whether to integrate with us.

A platform without published specs is a project. A platform with published specs is a product. The difference matters when your team is deciding where to invest engineering time.

We’re publishing because we want more partners building on top of the Developer Platform, and we know what developers look for. If you’ve integrated with Stripe or Twilio, you know the bar. We’re trying to meet it.

What’s next on the Developer Platform

OAS files are one piece of a bigger build. The Developer Platform also supports Model Context Protocol, or MCP, the emerging standard for connecting AI systems to data and tools. We have a growing library of MCP-compatible interfaces that let AI assistants like Claude or ChatGPT operate against our platform with the right permissions in place.

The roadmap from here is more endpoints, more partner integrations, deeper documentation tools, and continued investment in the kind of infrastructure developers expect from a serious platform.

If you build software in this space or run technology at a lender, the Developer Platform is worth a look. The OAS files are live today. So is the broader vision of what we’re building toward.