Rilibay is now an applied AI lab. Here is what that means.

Rilibay started as a small software studio with one rule. Build focused tools that do one job well. Our first product, Rilifi, runs live quizzes, polls and spin wheels. Creators, trainers and event teams in more than ten countries use it.

Today we’re narrowing that idea. Rilibay is now an applied AI lab. We build AI tools for specific industries. We build them with the people who run those industries. And we help teams use AI in ways that hold up in real work.

This note explains what that means, and what it doesn’t.

What “applied” means to us

A research lab asks what models can do. An applied lab asks a duller, more useful question. Which part of this job should a model do? And how will we know it’s doing it well?

That question shapes how we work:

• We start from the work, not the model. “Add AI to our business” isn’t a project. “Our coordinators spend two hours a day turning supplier emails into order sheets” is.
• We measure before we ship. Every AI feature gets a test set built from real examples we have permission to use. If we can’t say how often it’s right, it doesn’t ship.
• People stay in the loop. Our tools draft, sort, pull out data and suggest. People approve. A tool earns more freedom one task at a time, with proof.
• We like boring plumbing. Most of the value in an AI tool sits around the model. Think permissions, audit trails, retries, backups and a clean screen. That’s where we spend our time.

Why we build for one industry at a time

General AI assistants are impressive, and they keep getting better. But they’re general by design. The hard part of using AI at work is everything around the model.

It’s the words your industry uses. It’s the file formats and the odd cases. It’s the rules you must follow. It’s the five-step handoff between two teams that nobody ever wrote down.

A tool built for one industry learns that context once. Then nobody has to explain their job to a chatbot every morning. That’s the software we want to build: small, sharp and made for one kind of work.

The three things the lab does

1. We build our own products. Rilifi is build 001. Our next builds are AI tools for specific industries, and they’re in research now. We won’t announce them until they work. Join the waitlist to see them first.

2. We build with operators. Some of the best product ideas belong to people who run a business. They live with a broken process every day. Our co-build programme pairs what they know with our product, engineering and AI work. We agree the terms first.

3. We help teams adopt AI. Not every problem needs a new product. Sometimes the right answer is to add AI to the tools a team already uses. Sometimes it’s an honest “AI isn’t ready for this yet.”

What we’ll publish here

Lab Notes is where we write about the work. How we pick problems. How we test models for a task. What broke in production, and what we changed. We’ll be specific, and we’ll include the failures.

Do you have a task you think an AI tool should handle? Tell us about it. It’s the most useful message we can get.