Five days of work, done in three. Built in India.
Hospitals · Diagnostic labs · Imaging centres · Insurers & TPAs · Pharma
AI for Healthcare
Less retyping. Less searching. Faster decisions.
AI that reads forms, bills and drug safety reports, drafts notes and discharge files, answers patients on WhatsApp and flags what needs a closer look. Your team checks and decides.
Our team's work includes AI for two large US healthcare organisations.
The problem
Your best people are stuck on paperwork.
Hospitals
Beds stay blocked while insured discharges wait on summaries, bills and TPA files, so new admissions wait. Queries and deductions cut what you're paid: in FY2023-24, insurers cut or rejected ₹26,000 crore of health claims (IRDAI). And staff hours go on retyping.
Labs & imaging
Each retyped test request form risks an error, a repeat sample or an amended report: rework you pay for. Scarce radiologist time goes on reading old reports, so report turnaround slips.
Insurers & TPAs
IRDAI's clock runs on every cashless claim: 1 hour for a decision, 3 hours to authorise discharge. Miss the 3 hours and the insurer pays the extra hospital charges, while your teams still read every form and bill by hand.
Your teams should spend their time on care and decisions, not paperwork.
Many AI projects never get past the demo. We build the kind your teams use every day.
What we can build
Across hospital, lab, imaging, insurer and pharma work.
Your staff stay in control: the AI only follows rules they approve, they sign off where needed, and everything is logged.
Hospitals
Less waiting. More time for care.Faster discharge for insured patients
Discharge summary, final bill and TPA file prepared together for doctor sign-off, so beds free up sooner.
IRDAI: discharge authorised within 3 hrs of the hospital's request (India, 2024)
Pre-auth files, queries and deductions
Complete pre-auth files and full query replies, for fewer query rounds with TPAs and faster payment.
₹26,000 cr cut or rejected in FY2023-24, mostly on policy terms (IRDAI, India)
AI scribe for OPD and wards
Notes drafted from the consultation, so eyes stay on the patient. Doctors review every note.
US randomised trial: note time cut by up to 9.5%; we measure yours in the pilot
Sepsis and ward deterioration early warning
Flags patients on the ward who may be getting worse, from vitals and labs, for nurses to check.
Tested silently on your ward data first; may need CDSCO approval as a medical device
WhatsApp and voice patient assistant
Bookings, reports and reminders in Indian languages, all 24x7. Medical questions go to staff.
Text reminders cut no-shows from 21% to 15% (global meta-analysis, not AI-specific)
Diagnostic labs
Samples in faster.Test request forms and prescriptions
Printed and handwritten forms read into your LIS; staff confirm any unclear field.
About 1 in 10 handwritten requests had entry errors (peer-reviewed study, Australia)
Auto-verification with smart delta checks
Normal results auto-released on rules your lab director approves. Pathologists check the rest.
About 80% of HbA1c results auto-released, ~2 hrs sooner (China, hospital study)
Imaging centres
Reports out sooner.Structured reports from dictation
Dictation turned into your standard report template; the radiologist checks and signs.
AIIMS New Delhi used AI to structure 150 free-text CT reports (Indian study)
Prior-study summaries
Old reports and outside history summed up before the radiologist opens a follow-up scan.
AI histories named the diagnosis in 99.5% vs 89% of original requests (Canada, study)
CT bleed and stroke alerts
Head CTs that may show a bleed or stroke move to the top of the reading list.
Tested silently on your scans first; regulated as a medical device (CDSCO)
Insurers & TPAs
You decide on time, with the facts ready.Cashless decisions against IRDAI's clock
Pre-auth and discharge papers read, gaps flagged, so your team spends the clock deciding.
About 13% of pre-auths still missed the one-hour limit (Lok Sabha reply, India, 2025)
Fraud, waste and abuse detection
Flags unusual claims, hospital bills and documents, with reasons, for your investigators to check.
₹8,000–10,000 crore a year lost to fraud, waste and abuse (BCG estimate, India)
Pharma
Pharmacovigilance case intake and triage
Reports checked for validity and duplicates, coded, seriousness flagged. Your team verifies each case.
Required for new drugs under India's NDCT Rules 2019; CIOMS AI guidance, 2025
We can build AI across hospital, lab, imaging, insurer and pharma work. The two examples below show how we deliver; tagged rows reuse their technique.
Tagged: same technique as our US work, shown below.
Our work · 1 of 2
Reading incoming discharge summaries, 5× faster
For one of the largest kidney-care and dialysis networks in the US
Our team designed and built this system.
Before
Thousands of discharge summaries came in many formats. Nurses and clinicians extracted details by hand: slow, costly and error-prone.
What we built
Our AI reads each summary, whatever its format. A second, specialised AI handles unclear ones.
Why it matters in India
Indian hospitals and labs receive discharge summaries, referral letters and test request forms in every format, often scanned. The same approach can turn them into clean data for your HIS or LIS, so no one retypes them.
Result
5×
faster per summary: from ~25 to ~5 min, client-measured
~95%
of fields right first time
Much less manual work for nurses and clinicians.
Our work · 2 of 2
The whole patient story on one screen
For one of the largest and best-known teaching hospital groups in the US
Our team designed and built this system.
Before
Doctors spent a lot of time on scattered records. Standard screens showed plenty, but not what today's visit needed.
What we built
Notes, lab results, scan reports and outside records in one view, summed up for this speciality and this visit.
Why it matters in India
Patients bring folders of old reports from other hospitals and labs. Once scanned, the same approach can turn that folder into a one-page history.
Result
~2 hrs
saved per doctor per day, client-reported
10+
further uses built on the same base
Each fact links to its source, so it can be checked.
Client names withheld for confidentiality.
How we work
Three simple steps. You stay in control.
Pick one workflow
Choose one job, such as discharge files, a WhatsApp assistant or claim checks.
See it work
It runs on your own data. Output reaches your HIS, LIS or claims system via API, HL7 or files.
Go live
In daily use, we adjust its rules and settings, then hand it over or keep running it for you.
After the free test: a 6-week pilot, fixed price agreed up front.
For clinical alerts, scan AI and drug safety, we validate on your own data first, as ICMR's AI guidelines recommend.
How we handle your data
- We act as your data processor, under a signed agreement (DPDP Act).
- Patient data stays in your cloud or on your own servers.
- Your data never trains shared or third-party AI models.
- Your staff stay in control: the AI only follows rules they approve, they sign off where needed, and everything is logged.
- You own your data, the outputs and the code we write for you.
Output can be delivered as FHIR, the standard ABDM and NHCX build on.
Next step
Two easy ways to start
1 Test your documents
Free, no obligation. Send up to 10 of your hardest documents and see what our AI can do.
[email protected]2 Book a 30-minute call
Tell us where your teams lose the most time. We will say plainly if AI can help.
Email [email protected], or call or WhatsApp +91 90047 27477.
Who you’ll speak to
Sahil Kalra
Founder
Leads delivery: 8+ years building software, 3 in applied AI.
Akash Maggon
Co-founder
Builds assistants and workflow tools: 8+ years in software.
Himani Sharma
AI Engineer
Aman Lohan
AI Engineer
Founder-led AI engineers in India. Our name is our aim: five days of work, done in three.
Other industries