A database for decision models

Build apps that make their own decisions

Models are changing what software can do—and what people will expect it to do. Tasks that once needed a person’s judgement can now become part of the software itself. That opens up an entirely new range of applications that make their own decisions as they run.

Using AI to build legacy software is a bit like using a car to pull a carriage down a dirt road. We should be building highways instead.

See an example

Start your 7-day trial. No credit card required.

Understand the decisions behind your data.

In this software, models make decisions as it runs. Their answers become inputs to other decisions, change data, and lead to actions. Reading the code no longer tells you why something happened. Asking the model again may give you a different answer.

Your agent needs to see the decisions actually made—and follow their connections in both directions:

What led to this decision?

The data as it was at the time.
The code and instructions used.
Earlier decisions it relied on.
What the model saw and what it answered.

What followed from this decision?

The data changed using its answer.
Later decisions that relied on it.
Actions taken as a result, through many steps.

When a mistake comes to light, an explanation is only the beginning. Your agent needs to try a different approach, inspect what changed since, and record a carefully chosen correction.

Rotor keeps recorded work in one database.

Query historical data alongside current data.
Query dependencies your application explicitly records.
Retain files and application code as artifacts.
Record revisions alongside earlier versions.

See it in action: An AI-assisted job search

Find your next job with Rotor and your agent

An illustrative workflow

Your agent builds a shortlist. One recommendation looks wrong.

Follow a recommendation back to its sources.
See which later recommendations depended on it.
Try a revised approach without worrying about losing data.
Record revised recommendations and compare them with earlier decisions.

With Claude, Cursor, or Codex.