Clinical intelligence, under clinical control.
Aksh prepares worklists and drafts from the systems your clinic already uses, for patient access and clinical review. A named clinician decides every step, and every decision is recorded. It never acts on its own.
Cancelled-slot fillCGM monitoringPatient app
Clinical AI fails in two predictable ways.
Aksh is one system beside the EHR, so every capability follows the same rules for safety, audit and consent.
Point tools that don’t compose
Separate risk, coding and coaching apps, each with its own trust and safety boundary. The clinician becomes the integration layer.
Automation that oversteps
Auto-writes to the record, auto-dosing, diagnostic claims, with nobody accountable in the loop.
Aksh’s answer
- One supervisor: runs every component in a fixed order that can be inspected
- One audit trail: who drafted, who decided, what changed
- One consent model: shared by every module and role
- One rule: nothing reaches a record or prescription without a human decision
Two products on one foundation.
Patient access and clinical review share sign-in, consent, connections and audit, so both follow the same rules. Each item shows where it is today.
- Production
- Deployed and verified in production
- In validation
- Built; not yet released for real patients
- Synthetic demo
- Works on synthetic data only
- Planned
- Not built yet
Patient access
Fill freed appointments from your waiting list, with staff in charge
- Cancelled-slot fillFreed slots matched to the waiting list with reasons; staff choose and contact; the slot counts only when the schedule confirms the booking.Production
- Hand-offsAnything that needs a person goes to a named queue with a reason and a due time.Production
- Weekly resultsFill rate, staff time and issues each week, against a baseline locked before go-live, with sign-off.Production
- Patient requestsRoutine questions and scheduling requests handled by an assistant, passed to staff when needed.Synthetic demo
- Recall and missed-appointment outreachOutreach to patients who missed or are overdue for appointments, using approved wording.Planned
Clinical review
Worklists and drafts for clinicians to accept, edit or reject
- Glucose and CGM monitoringCGM and shared readings as one worklist, low glucose firstProduction
- Clinical review queueEvery draft in one queue; a clinician accepts, edits or rejects it on the server, with the decision and audit record saved together.In validation
- Notes and coding draftsStructured consultation notes and coding drafts for clinician review.In validation
- Medication safety and referralsInteraction and allergy checks, risk changes and referral drafts, each held for a clinician.In validation
Shared foundations
Both products use the same sign-in, connections, consent and auditFrom data in to a recorded decision.
The clinician decision sits in the middle. Nothing after it happens without that decision.
Data in
Schedules, readings, CGM data, lab results and what patients share
Organise
Matched to the right patient, checked, and gaps flagged
Prepare
Worklists, ranked options and drafts, each with its reasons
Check
Safety rules applied before anything reaches a person
Clinician decides
Accept, edit or reject; some decisions need a named role
Act
Your staff act through the systems and channels you already use
Record
Every decision kept with who made it and when
Review
Weekly results against a baseline set before go-live
Diabetes care, one worklist at a time.
Glucose monitoring is available today. The rest of the module adds to the same worklist and review queue, with the same safeguards.
- Production
- Deployed and verified in production
- In validation
- Built; not yet released for real patients
Glucose and CGM monitoring
ProductionTime in range, low and high glucose, variability and daily patterns, as a worklist
Low glucose first; the clinician decides the response
Readings from patients
ProductionGlucose, blood pressure and weight shared from the patient app
Marked as patient-reported until a clinician checks them
HbA1c trends
In validationLatest value and trend over time, in the organisation's units.
Clinician review before any change to the care plan
Medication safety and referrals
In validationInteraction, allergy and low-glucose risk flags for insulin and other medicines
Explains; never sets or changes a dose
Complication checks
In validationKidney, eye, foot, heart and liver checks that are due or overdue
Referrals need a clinician's confirmation
Lifestyle support
In validationNutrition, activity and habit talking points for consultations
Concerns raised by patients go to a person
Ready for clinics today.
These workflows are in production and ready for a clinic to go live. Each one starts in shadow mode.
Cancelled-slot fill
Freed appointments offered to the patients who need them most, with the reasons. Staff decide every contact.
CGM monitoring
Time in range and consensus targets as a worklist, low glucose first.
Patient app
Readings and CGM uploads shared with the care team, on the patient’s terms.
Measured results
Weekly review against a baseline locked before go-live.
Every output is a draft until a named clinician decides.
- Accept, edit or reject before anything can affect the record.
- Every decision is attributed to a named reviewer and audited.
- Missing data, like an overdue eye-screening date, is listed rather than guessed.
- No autonomous diagnosis, prescribing or treatment.
Mandatory human gates
- 1Urgent concernsResponsible clinician confirms
- 2ReferralsResponsible clinician confirms
- 3Medication decisionsPrescriber confirms
- 4Every other draftAccepted, edited or rejected by a named clinician
Enforced on the server, not just hidden in the interface.
What Aksh will not do.
These limits are part of the design. We say clearly what is available today and what is still in development.
- No clinical decision without a clinician
- No automatic prescribing or dose changes
- Not a replacement for your records system
- Nothing written to your records without approval
- No silent changes to how Aksh works
- No claims of certification we do not hold
- No output without its reasons
- No patient data sent to AI models in production
From first conversation to live clinic, one step at a time.
- 01
Scope one workflow
Pick the diabetes workflow to start with, name the reviewers and set the governance boundaries.
- 02
Set up your clinic
Invite your team, check registrations, load patients and the waiting list.
- 03
Complete go-live
Data processing agreement, privacy assessment, clinical safety lead and a locked baseline.
- 04
Widen on evidence
Start in shadow mode; add workflows only when the evidence supports it.
Clinical intelligence should help clinicians think, not replace them.
Tell us about your organisation. Clinics go live one workflow at a time.