Describe the work
Plain English for the data work and the modeling goal.
Epiphany deploys the production pipeline, opens Data Lab on the output, and one-clicks training, evaluation, and deploy in ML Lab.
Pipeline, lab, and model as one product
Not a bolt-on notebook. Data Lab and ML Lab consume the same pipeline outputs and the same AI agents.
ML Lab
Same agents that built the pipeline scan targets, train a baseline, iterate experiments, and deploy the model back onto the job.
- One-click from pipeline output
- Classification & regression templates
- Train → evaluate → deploy
Data Lab
Interactive analysis on live pipeline outputs: profiling, anomalies, quality, visualizations, and NL questions — fed by the same AI crew.
- Profile & quality scorecards
- Anomaly detection
- Promote findings back to the pipeline
Data Quality
Automated profiling, anomaly detection, and quality scorecards on every pipeline run
- Column profiling & distributions
- Anomaly detection
- Quality scorecards
API & CLI Access
Full REST API and a Go CLI for terminal-first workflows
- REST API with Bearer auth
- Go CLI for plan → deploy
- --json flag for scripting
Cost Analytics
Transparent cloud costs and usage tracking per pipeline
- Real-time tracking
- Budget alerts
- Cost optimization
Compliance & Security
SOC 2 Type II, GDPR, HIPAA-ready with encrypted credentials and RBAC
- SOC 2 Type II
- GDPR compliant
- Encrypted credentials
From English to production ML
Describe the data work and the modeling goal. Get a pipeline, analysis outputs, and a trained model.
Step 1 Screenshot
Query Input Interface
Describe the work and the goal
Sources, transforms, and what you want to learn or predict — in one sentence. No notebooks to stand up first.
Step 2 Screenshot
AI Building Pipeline
Agents build the pipeline and the lab
Planner, ingestion, and transform agents deploy ETL. Lab agents profile the output, flag anomalies, and set up experiments.
Step 3 Screenshot
Ready to Deploy
Train, evaluate, deploy in one click
Templates take pipeline outputs into a baseline job, then evaluation and model attach on the same pipeline. Iterate without leaving the product.
From concept to production in seconds
Work with everything
Your entire data stack, unified
Built for analysts who also ship models
One description. Production pipeline, interactive lab, trained model.