New to GeneFlow

Audience: anyone who doesn't fit a more specific role above, or wants a high-level on-ramp before diving into a role workbook.

What is GeneFlow?

GeneFlow is GeneData's ML & GeneAI lifecycle platform. It speaks MLflow's wire protocol (so existing mlflow.start_run() code works) and adds the production layer: prompts as first-class entities, real-time collab, cost-per-run, hash-chained audit, K8s-native runners, model serving with drift detection.

In one sentence: GeneFlow is the open-standards lifecycle platform for both classical ML and GeneAI workflows.

When you'll use it

For everyone else, see the full workbook index.

Five-minute tour

1. Open /dashboard/geneflow/ml
2. Experiments tab — runs grouped by experiment, sorted by recency
3. Models tab — registered models with version status pills
4. Endpoints tab — live model-serving with QPS / p95 / drift status
5. Prompts tab — versioned prompt registry

First commands

pip install geneflow
export GENEFLOW_TRACKING_URI=https://api.genedata.io
export GENEDATA_PAT=$(genedata auth token)
export GENEFLOW_TENANT_ID=ACME

gfctl experiments list
gfctl endpoints list
gfctl prompts list

Five-minute first run

import geneflow

geneflow.set_tracking_uri("https://api.genedata.io")
geneflow.set_tenant("ACME")

with geneflow.start_run(experiment_name="my-first-run") as run:
    geneflow.log_param("greeting", "hello")
    geneflow.log_metric("answer", 42)
    print(f"Run ID: {run.run_id}")

Now look at it in the UI: /dashboard/geneflow/ml → filter by your experiment name.

Vocabulary

Help

• Slack: #geneflow
• Docs index: ../README.md
• Role workbooks: ./README.md
• API reference: ../api-reference.md