What is predictAI?
The Prediction OS — one platform that turns your company's data into answers for every decision that depends on what happens next.
predictAI is a Prediction OS: an operating system for every decision your company makes about the future. How much will we sell next month? When will demand spike? Which metric is about to drift? What happens if we raise the price? Every one of those is a prediction problem — and predictAI is where they all live.
Your data is ingested from wherever it already sits — databases, SaaS tools, files, event streams. The platform learns what actually drives your outcomes, trains and compares models, keeps the best one live, and answers over a simple HTTP API — continuously, and getting better as new data arrives.
Everything is programmable. The whole platform is one API, and these docs cover all of it.
The loop
Everything in predictAI fits one loop:
- Ingest — connect a database or SaaS tool, upload a file, or send events over HTTP. Your data flows in from wherever it lives.
- Signals — ingested data becomes named, analyzable signals: metrics, measurements, and the events that move them.
- Model — pick or build a model: your own architecture, a pretrained foundation model, or one you trained elsewhere.
- Train — a pipeline trains the model on your data, on a schedule or on demand.
- Deploy — put the best result into live serving.
- Predict — call one URL and get answers.
- Goals — or skip steps 3–5 entirely: state the decision you need answered, and the platform discovers what drives it, runs a model tournament, and keeps a champion serving for you.
Quickstart
Token → send data → create a goal → get a prediction. Five minutes.
Core concepts
Workspaces, signals, models vs pipelines, deployments and goals — the mental model.
API Reference
Base URLs, headers, conventions, and every endpoint in one place.
Goals
The headline feature: outcome-driven forecasting on autopilot.
Where things live
| You want to… | Go to |
|---|---|
| Bring data in | Ingestion |
| Browse and analyze your data | Signals |
| Build the table a model trains on | Segments |
| Pick or build a trainable model | Models |
| Set up training workflows | Pipelines |
| Inspect training runs and results | Trainings |
| Get predictions | Inference |
| Manage what's live | Deployments |
| Automate the whole thing | Goals |
| Understand costs and quotas | Billing |
| Manage tenancy and members | Workspaces |
| Manage tokens and your profile | Account |
| Get told when things happen | Notifications · Realtime |