FOUNDRY

by Akuru

Incorrect password


AKURU FOUNDRY

The compliant platform where
medical AI gets built and deployed.

Clinicians are already building their own tools with AI. Foundry is the Australian-sovereign, clinically-governed platform that lets them deploy those tools properly, and sell them.

The Problem

Clinicians are already building their own medical tools. They have nowhere safe to run them.

The code was never the hard part, and AI has made it trivial. The hard part is that nothing built this way is appropriate to run: patient data ends up on unvetted platforms, with no residency, no consent capture, no audit, and real professional risk attached to every deploy.

Data

Patient-identifiable data on consumer platforms with no residency guarantee.

Registration

One privacy breach or bad output, and it is the clinician's AHPRA registration on the line.

Nowhere to land

No compliant deploy target exists for a solo clinician-builder. So the tools stay stuck on a laptop, or ship unsafely.

Why Now

Soon, building an app will be as common as making a spreadsheet.

The first wave of healthcare AI built tools. The next wave lets everyone build their own. The curve is not coming, it is here: in six weeks, three separate frontier models each leapfrogged the last.

9 June 2026

Fable

New frontier release

9 July 2026

GPT-5.6 Sol

Jumps in front

16 July 2026

Kimi K3

Jumps in front again

The leapfrogging is the thesis: if the frontier reshuffles every month, betting on the model is a losing game. The durable position is the substrate the models plug into.

The Thesis

The coding gets free. The plumbing becomes the product.

Foundry is the compliant platform-as-a-service for medical AI, built to be deployed to by AI coding agents, not humans clicking consoles. But the deploy target is table stakes. The real value is the shelf of medical primitives no generic platform has and no model lab will ever build.

Concede

The software layer

Any AI agent will eventually rebuild any feature we ship. We do not fight that. We assume it.

Own

The substrate

Compliance, integration, contractual reach, and the clinical primitives: the things an agent structurally cannot do for a solo builder.

Evidence of Pull

Our own customers are already building their own tools, and coming to us for help doing it.

This is not a hypothesis. Clinicians who use our products are vibe-coding their own software right now, some trying to rebuild i-scribe itself, and asking us how. The demand arrives as inbound help requests: they have not left us, they want to build with us.

7
Builders signed up
9
Apps already built
24 hrs
From first enquiry

"They are trying to rebuild our own product themselves, and asking us for advice on how to do it."

Seven clinician-builders with nine finished apps signed up wanting to deploy on Foundry within a day of us asking. Demand this warm, this early, without a product to show, is the whole thesis in one number.

What Foundry Is

The parts you should not build alone are already on the shelf.

Hardened in our own products, in real clinical use. A builder's agent imports them like any other library, except these have already passed clinical review. Vercel does not have them. Anthropic and OpenAI never will: they are jurisdiction-specific, contract-encumbered, and clinically hardened.

foundry/billingEclipse & Medicare billing. MBS items, claiming, patient payments, GST handling correct from the start.
foundry/scribeScribe-tuned models. The clinical scribing and history-taking engines behind our own products, as building blocks.
foundry/de-idDe-identification as a service. Cannot be skipped on AI calls. Nobody can safely vibe-code this.
foundry/emrEMR extraction. The way in and out of Australia's walled-garden EMRs. See appendix.
foundry/ahpraAHPRA verification. Confirm users are registered clinicians before they touch anything clinical.
foundry/consentPatient consent capture. Built into the platform, not bolted on per app.
foundry/prescribeElectronic prescribing. A heavyweight regulated primitive with ERx conformance nobody vibe-codes: a moat in one line.
foundry/commsSMS & email sending. Compliant patient-comms plumbing every app needs and nobody wants to rebuild.

The Moat

An AI agent can write any feature. It cannot sign a contract.

Assume the labs win agents, models, and even hosting. Here is what they will never do: support your ISO 27001 application, walk into a hospital procurement meeting as your compliance co-signatory, or sign billing agreements with every Australian health insurer. Concede the software layer. Win the contractual last mile.

Contractual reach

Insurer billing agreements, hospital procurement, ISO 27001 inheritance. A coding agent cannot sign any of them.

Regulatory depth

Privacy Act, TGA SaMD boundaries, MBS, AHPRA. Local, unglamorous, exactly what US-native players and model labs will not build.

Instant assist

Because the code and environments live with us, our engineers can jump into any builder's app and fix it in minutes. Leaving Foundry means being alone with your code again.

 CompliantSpecialist primitivesSigns the contracts
Generic platforms (Vercel, Replit)nonono
Model labs (Anthropic, OpenAI)partialnono
Dev agencyslow / costlynono
Akuru Foundryyesyesyes

Business Model

Build for yourself for next to nothing. We earn when you start selling.

Usage-based billing, kept deliberately cheap for personal-use builders. The point is to get everyone in making things for themselves: that is top-of-funnel, not revenue. The money comes from the builders who sell, and from clinics and enterprises running apps at scale.

Loss-leader by design

Personal use

As cheap as we can make it. Every clinician tinkering for themselves is a future seller and a future channel.

Where the revenue is

Sellers & enterprise

Marketplace rev-share, a 15% block take-rate (comfortably under the 30% app-store anchor), and clinic / enterprise usage.

Services on top

FDE & support

Forward-deployed engineers for enterprise, and "if it breaks mid-clinic, we fix it now" support plans.

The 15% take-rate is a recruitment line in itself: "fairer than Apple" writes itself when courting the specialists who publish the underlying components.

Go To Market

Walk in with an idea. Walk out with a live app.

Half-day build sessions run at specialty colleges, GP conferences, and practice-manager events. The promise, stated exactly: arrive with an idea, leave with your own app vibe-coded, deployed, live, and ready for clinical use, the same day.

  1. 01Activation is the event. A room of prospects becomes a room of live tenants in an afternoon. No follow-up leakage.
  2. 02They build what they actually want. Their own itch, scratched, not a canned tutorial. That is a tool they open on Monday.
  3. 03The badge is the marketing. Every shared app carries "Built with Akuru Foundry": the "Sent from my iPhone" of the platform.
  4. 04CPD accreditation is the unlock. Australian doctors always need CPD hours. Accreditation turns attendance from a favour into something they need.

Why Us

We already live in 2028 internally. This raise sells the time machine.

Around 70% of Akuru's codebase is AI-generated, and agents run roughly half of our internal operations. This is not a pivot: it is Akuru bringing the internal tools we already run every day to the world. Continuity, not departure.

~70%
AI-generated codebase
3
Clinical products in production
ISO 27001
Certified posture, inherited

We start with tools for clinicians. We end as the platform medicine is built on.

The closing line only lands because the opening is narrow and evidenced: win the specialist builder first, become the substrate specialist care is delivered through.

The Team

Nimesh Piyatissa
CEO & Co-Founder
Emergency physician & founder

Leads strategy, growth, and partnerships across healthcare providers, specialist groups, research institutions, and professional bodies. Combines clinical credibility, commercial vision, and systems thinking to position Akuru as trusted infrastructure for specialist care.

Chandrashan Perera
CTO & Co-Founder
Ophthalmologist & AI engineer

Leads Akuru's technology and product architecture. Builds the intelligence layer that turns real clinical workflows into safe, scalable software for the future of care.

The Raise

$1M buys exactly the part agents can't do.

This is not platform-build money: the platform is our internal infrastructure being made external. The seed buys the integration and contractual last mile. Every dollar builds more integrations, directly or through the contracts that unlock them.

4 hires

Backend / integration devs

The integrations pipeline as a production line: EMR extraction, billing, connectors.

1 hire

Enterprise sales BDM

Dual-purpose: the contracts pipeline that unlocks integrations also feeds the FDE / enterprise revenue line.

$200k

Hackathon marketing

Co-branded builder marketing and CPD-accredited events. The funded channel. No paid ads.

The Ask

We're raising $1M on a $50M cap.

A SAFE note, priced to reflect what Foundry already is: Akuru's proven internal infrastructure, made external, with warm demand on day one. Download the note, or book a call and we'll walk you through it.

$1M
Raising
$50M
Valuation cap
SAFE
Instrument

Reviewed the note and ready to talk? Book a call above, or send the signed SAFE to chandra@akuru.com.au.

Appendix

For the reader who wants the deeper mechanics. Keep scrolling.

A1 · What You Inherit

Everything you would rather not think about, already in place.

The moment an app lands on Foundry it inherits the posture that takes healthcare companies years to build. None of it required configuration. None of it can be switched off by accident.

Data residencyAustralian-hosted. Data never leaves.
Privacy Act alignmentBy default, not per-app.
EncryptionAt rest and in transit.
Managed data layerFoundry provisions the database. Dev / staging / prod separated. The coding agent can only ever touch dev data: your AI can't leak what it can't reach.
Audit loggingEvery access, always.
Assurance pipelineSecurity, PII-leak, prompt-injection and clinical-safety checks on every deploy. Findings are tiered: auto-fix or explain.

A2 · EMR Liberation Layer

Australian EMRs are walled gardens. We are the way in and out.

A clinician-built app is useless without patient context, and Best Practice, Medical Director, Genie and Zedmed have no clean APIs. Whoever solves data-in and data-out becomes the layer every app depends on. Four extraction models, in ascending difficulty:

  1. 01Document extraction. Referrals, discharge summaries, pathology PDFs to structured data. Highest volume, cleanest legal position, best training-data story. Ships first.
  2. 02Free-text note extraction. Problems, meds, allergies out of GP progress-note chaos. i-scribe already runs this in reverse.
  3. 03HL7 / legacy normalisation. Pathology feeds and CDA to clean JSON. Unsexy, universal, nobody builds it twice.
  4. 04Screen-level extraction. When there is no API, read the EMR the way a human does. The universal adapter.

A3 · The Marketplace

Three sides, one flywheel.

Users bring revenue and builders bring apps, but block-providers bring compounding capability Akuru never has to build. Specialists with great models cannot commercialise today: Foundry supplies compliance, billing, distribution and an audience, for a take-rate.

Tier 1 · Users

Clinicians & practices

Install and use finished apps. Pay per app / subscription.

Tier 2 · Builders

Vibe-coding clinicians

Build from scratch or from templates. Free personally, earn via the marketplace.

Tier 3 · Block providers

Specialists & ML teams

Publish components (a fundus-grading model, a billing API) at a self-set rate. Foundry takes 15%.

Blocks make builders more capable, which makes more and better apps, which brings more users, which drives more demand for blocks. Three-sided flywheels are hard to bootstrap and nearly impossible to dislodge: the strongest answer to "what if a hyperscaler copies you." They would be copying an ecosystem, not a feature.

A4 · Market

The software market is large. The care-delivery market is much larger.

Foundry enters through specialist workflow software, but the real opportunity is owning the build-and-deploy layer under all of medical software.

Where Foundry is headed

Medical build & deploy infrastructure

The compliant substrate every clinical app runs on

substrate

Platform

Specialist workflow platform

Intake, documentation, coordination, billing

A$8.19B

Entry point

Documentation software

Clinical notes and transcription

~A$1B

Entry point → platform → infrastructure.

A5 · Rollout & Metrics

Ten committed apps today. Five weekly-active apps is the milestone.

Integration-gated cohorting is the organising principle. Nothing is turned away: apps that cannot yet deploy are queued, and the queue is the integration roadmap.

  1. 0Phase 0. Outreach for already-built apps looking for a home. Target: 10 committed apps in the pipeline. (7 builders / 9 apps already, in 24 hours.)
  2. 1Phase 1. Deploy the apps needing no integrations, then those needing integrations we already have. Battle-test the platform. Success = 5 apps published and weekly-active.
  3. 2Phase 2. Build the core integration set, ranked by real early-access demand. Waitlist wide-open, deployment gated by integration coverage.
  4. 3Phase 3. The loud launch: conference hackathons, CPD events, the collab-post engine. Only after Phase 2 is live.

Primary metric

Apps deployed

Cumulative apps live on the platform. Builders come.

Primary metric

Weekly Active Apps

Apps with more than 4 uses in a week. Apps get used, not just shipped.

Primary metric

Active users

Distinct weekly users. Users, not builders: the network beyond the makers.