AI · Studio
Build AI with any model, and use it in any application.
AI Studio is where your team builds its AI without writing code: connect any model, add your documents, write and test prompts, build pipelines that answer with citations, give agents tools, and check every change before it goes live. What you build is a plain definition you can version, release, roll back and call from any system.
Everything AI needs, in one workspace
Models and profiles
Any OpenAI-compatible model, with your own key. Everything else refers to a named profile, so you can switch models without touching anything else.
Knowledge bases and embeddings
Turn documents into passages and search them by meaning and by exact words. Use any embedding model.
Prompts
Versions, labels such as prod, and a playground that runs on your real models.
RAG pipelines
Answers taken from your knowledge, with the source shown and every step visible.
AI agents
A model with tools and knowledge, at an autonomy level you choose, with approvals and a trace.
Evaluations and guardrails
Test questions and scorers before a release, and checks for personal data and prompt injection.
The real screens
These are AI Studio itself, running on a local model with real data.
From idea to a working feature
Connect a model, or use one the platform offers, and give it a profile name such as default.
Add the documents your answers should come from and test what a question finds.
Write a prompt, try it in the playground, and label the version you trust as prod.
Build a pipeline that answers from your knowledge with citations, and watch each step.
Add test questions and run them before you move a label.
Call it from a workflow step, a plugin, or any application through an OpenAI-compatible endpoint.
Use it from anywhere
Workflows
Add AI steps to a business workflow; the workflow keeps control of the path and the approvals.
Plugins
Ship prompts, pipelines, agents and knowledge inside a plugin so every tenant that installs it gets them.
Any application
An OpenAI-compatible endpoint with app API keys: change the base address and key in any OpenAI SDK.
Your data stays yours
Each tenant's AI data, including documents, vectors, prompts and run history, lives in that tenant's own database and is never mixed with another tenant's. Provider keys are stored as secrets and never shown again. The model, not our platform, decides what to say; the platform decides what it may do.
What to know today
Knowledge bases take pasted text, text files and web pages; PDF and Word files are not supported yet. Answers arrive complete rather than streamed. Prompt caching is not used yet. Budgets and alerts are not built: usage and cost are shown, not enforced.
See AI Studio on your own documents.
Tell us the questions your teams keep asking and we will show you how it answers them.