Skip to main content
    All work

    Venture 01 · In-house · In development

    Sukham — one calm operating flow for the everyday clinic.

    A clickable prototype for small Indian OPD clinics. It connects simulated ABHA-based intake, consultation, bilingual e-prescriptions, GST and UPI billing, WhatsApp reminders, a consented patient record and clinic-owner visibility. A source-linked patient context model is the next phase.

    Interactive prototypeSimulated integrationsAI context layer in R&D
    View the product story

    The problem

    Departmental handoffs, not one visit

    Registration, consultation, prescribing, billing and follow-up run on different books, apps and staff. Context is lost between them.

    The thesis

    One visit, one thread

    A visit should move as one thread. The clinician stays with the patient, and the owner sees the whole day at a glance.

    The workflow

    Intake → consultation → prescription → payment → consented record.

    1. 01

      ABHA intake (simulated)

      Scan-and-share style intake collects patient context before the visit begins.

    2. 02

      Queue → clinician workspace

      One thread carries context, deltas since the last visit and missing-information flags into the consultation.

    3. 03

      Bilingual e-prescription

      A draft is generated for the clinician to review; nothing leaves the room without approval.

    4. 04

      GST + UPI billing (simulated)

      Billing and receipts come from the same visit thread, not a separate ledger.

    5. 05

      WhatsApp reminders (simulated)

      Follow-ups are queued from the visit record with explicit patient consent.

    6. 06

      Consented patient record

      The record is written back to a consented store, owned by the patient, visible to the clinic owner.

    All external integrations shown above are simulated in the current prototype.

    The AI direction

    A source-linked patient context model

    • · Source-linked pre-visit briefs
    • · Changes since the previous visit
    • · Missing-information flags
    • · Routine-documentation drafts

    We say "patient context model", not "digital twin", because the prototype is a simulated workflow rather than a predictive patient representation.

    Human control

    The clinician approves every clinical output

    Drafts assist; they do not decide. Every prescription, summary and downstream instruction is reviewed before it leaves the room.

    Trust

    Safety by construction

    • · Provenance on every fact, summary and inference
    • · Uncertainty is shown to the clinician, not hidden
    • · Consent gates every external integration
    • · Role-based access for owner, clinician and front desk
    • · Audit trail on every write
    • · Clear separation of facts, summaries and inferences

    Evidence

    Measured before benefits are published

    • · Time from arrival to prescription
    • · Corrections applied by the clinician to drafts
    • · Draft acceptance rate
    • · Offline recovery after a network drop
    • · Subgroup performance before any benefit is published

    Building something similar in a regulated field?

    Bring us the problem that crosses product, AI and operations. We start with a diagnosis.

    Talk to a senior partner