ekascribe

On-prem, self-hostable AI scribe: record a consultation → transcribe → structured clinical note → edit → print/copy.

Run locally

make start

Builds everything, initializes the DB, and starts the stack in Docker. App + API at http://localhost:8000. First run creates .env from .env.example — add SARVAM_API_KEY and your LLM key. Stop with make down.

Run in prod

On the VM, set keys, ENV=prod, and SELF_URL=https://<your-domain> in .env, then:

make start-prod

Serves on container port 8000 — front it with your HTTPS proxy (mic needs a secure context).

Run on Kubernetes

./deploy/push.sh <tag>          # build + push images
kubectl apply -f deploy/k8s/    # deploy to namespace eka-care
kubectl -n eka-care port-forward svc/ekascribe-api 8000:8000

Secrets and details: deploy/k8s/README.md.

Docs

• docs/architecture.md — how the pieces fit
• CONTRIBUTING.md — dev workflow