Case study · Our product

Amara Health: one health memory from every lab report

Amara Health turns lab reports from any lab, year and format into one health history you own. It lines up every value, shows what changed since the last report, and links every answer back to the original document. Built and run by MetaNova AI.

health-memory · illustrative

Lab history · common vocabulary

202420252026
HbA1c %5.96.15.8
LDL cholesterol mg/dL132118104
Vitamin D ng/mL182431

Every value links to its source report

3 reports merged, units normalised ✓

At a glance

The product in brief

  • What it is: an AI health memory layer for lab reports, for individuals, families, labs, hospitals, clinics and employers
  • Status: in public beta since July 2026, with a hospital pilot under way and ABDM sandbox milestones M1 and M2 complete
  • Platform: web, with records held in Indian cloud infrastructure
  • Languages: English, Hindi, Telugu and Tamil
  • Principle: Amara does not diagnose. It shows what changed and leaves the meaning to your doctor

The problem

Why is a health history so hard to use?

Every lab lays out its report differently. Page four of one lab's report is page six of another's. The same test appears under different names, in different units, with different reference ranges. The reports themselves land wherever they were sent: WhatsApp, email, a drawer, a camera roll.

So when someone sits down with a doctor, the history that would help most is the history that is hardest to find. Twenty pages to spot what is out of range. A second report that never gets opened. Last year's value that nobody lines up against this year's.

Most tools solve a smaller problem. They make one lab report easier to read. Very few connect the reports a person already has into one history and show what changed between them.

What we set out to do

The brief we gave ourselves

  • Accept a report from any lab or hospital, with no integration needed on the lab's side
  • Turn many differently written reports into one comparable record
  • Make the comparison the main event: what changed between the last report and this one
  • Keep every statement tied to its source, so nothing has to be taken on trust
  • Say plainly what the product does not do, and never drift into diagnosis

How we approached it

Memory first, explanation second

01

Treat the reports as the product

The value is not a clever summary of one report. It is the connected history across many. We designed the data model around a person, their reports over time and the individual values inside them, and built the interface on top of that.

02

Normalise before you explain

Before any value can be compared it has to mean the same thing. We built a common vocabulary of test names, converted units and kept each report's own printed reference range alongside the value, so a flag always reflects the range on that report.

03

Show it before asking for an account

The live demo on the site reads two sample reports, or two the visitor adds, and shows the comparison in the browser. A visitor sees the product work in about a minute, and files they add are parsed locally and are never uploaded.

04

Draw the safety line early

Product language, screens and answers were written around one rule: show what the report says and what changed, point to the original, and send medical meaning back to a doctor.

What we built

The product, feature by feature

01

Reads any lab report

Reports from different labs and in different layouts are read value by value. Each reading keeps a link to the original document, which stays one tap away.

02

One common vocabulary

Test names such as HbA1c, Fasting Glucose, LDL or Vitamin B12 are mapped to a shared set, with alternate names handled, so values from different labs line up in the same row.

03

What Changed

Pick two reports and Amara lines up every shared value, shows the previous and current reading with dates, the change in absolute and percentage terms, and whether each value sits inside or outside the printed range. A diabetes follow-up, a thyroid check and a kidney watch across two different labs are all the same screen.

04

Trends over time

Values are plotted across every report, with a table view for people who prefer numbers. Three years of follow-up from several labs becomes a single line instead of a pile of PDFs.

05

Ask your reports

Questions can be asked in English, Hindi, Telugu or Tamil. Answers are drawn from the person's own reports and each one links back to the report it came from.

06

Family accounts

A separate history for each person in a family, kept apart, so a parent's reports from one lab and a child's from another are ready when either sees a doctor.

07

Sharing with consent

The owner decides what a clinic or doctor can see and for how long. Access can be time limited, for example 24 hours, or kept until it is revoked, and can be revoked at any time.

08

Co-branded for partners

Labs, hospitals, clinics, telehealth services and employers can offer Amara under their own name. It sits after the report or the consult, so no integration with a lab or hospital system is needed to start.

09

A reference for 48 common values

A plain language guide explains 48 values that appear on common Indian lab reports: what each one is, a typical adult range and why a doctor looks at it, with a note that ranges vary and the printed range on the person's own report is the one to use.

Design decisions

The choices behind the product

01

Privacy as a product feature

Records are encrypted at rest and in transit, held in Indian cloud infrastructure, never sold and shared only by the owner's choice. The product is built for India's data protection law, and the privacy page says so in plain words.

02

Every answer has a source

An assistant that cannot show where an answer came from is asking to be trusted blindly. Amara links each answer to the exact report, which is also what makes it useful to a doctor in a sixty second consult.

03

Honest about its limits

The tagline is that Amara does not diagnose, you do. Disclaimers sit beside the numbers, and the product states that it supports lab reports today, with prescriptions and discharge summaries to follow.

04

Priced simply

A free tier with two reports and the full comparison lets people try the real thing. The Individual plan is ₹999 a year for one person with unlimited reports, a Family plan for up to six people is on its way, and partners can start at zero cost.

05

Built to be found and checked

The product site includes a comparison of product categories, an FAQ and a full reference page for each of the 48 values, so the claims can be read and tested by anyone before they sign up.

Under the hood

How it is built

  • Document reading pipeline that extracts values, units and ranges from differently formatted lab reports
  • Test name mapping and unit normalisation into a common vocabulary
  • Time series storage per person, per test, with a link back to the source document
  • Retrieval over a person's own reports, so questions are answered from their data and cite it
  • Multilingual question and answer in English, Hindi, Telugu and Tamil
  • Encryption at rest and in transit, with Indian data residency and consent based sharing
  • In-browser parsing for the public demo so visitor files never leave the device

Where it stands

What has happened so far

  • In public beta since July 2026
  • ABDM sandbox milestones M1 and M2 complete, with M3 evaluation active
  • A pilot under way with a 920 bed hospital in Visakhapatnam
  • Recognised as a startup by DPIIT
  • A live public demo that shows the full comparison on two reports without an account

What comes next

The roadmap

  • The Family plan for up to six people
  • More record types: prescriptions, discharge summaries and other medical records
  • Deeper ABDM integration following the sandbox milestones
  • More partner deployments for labs, clinics, hospitals and employers

Have a workflow like this? Tell us how your team works today.

Book a 20 minute workflow review