What is DinoAI?

DinoAI is Paradime’s AI data engineering layer. It’s not one chatbot; it shows up in three forms, each suited to a different kind of work. Knowing which is which makes the rest of the AI docs much easier to navigate.

​

The three forms

​

Copilot: AI in your editor

The copilot works with you inside the Code IDE. You prompt it, it reads your project and warehouse schema, and it writes or fixes code with you in the loop: generating sources and models, writing documentation, fixing failing commands, and drafting pull request descriptions. Use it when you’re actively developing and want to move faster.

​

Autonomous agents: AI in the background

Agents run in the Agent app without you watching. You give an agent a task, it works on your project in its own environment, and it comes back with results, like a branch and pull request. Bolt can also trigger agents automatically, for example to diagnose a failed production run. Use them when the task can run without you, like backfilling docs or healing a broken pipeline.

​

Programmable agents: AI you configure

Programmable agents are recipes you define and trigger on your terms, through schedules, webhooks, or CI. The guides include ready-made patterns like a pipeline healer, a test maintainer, and a PR reviewer. Use them when you want repeatable, automated AI workflows wired into your stack.

​

What makes DinoAI different

DinoAI is warehouse-aware and project-aware. It reads your dbt™ project, your schema, and your conventions (via .dinorules) rather than guessing from the prompt alone. That context is why it can generate a correct sources.yml from your actual warehouse metadata instead of a plausible-looking one.

​

Where to go next