HLTH 2026 · Nov 15–18
Book a slot →A data & AI team only works if it behaves like your own: a straight answer on whether to build before how, an architect on the first call, a demo every two weeks, and everything handed over as yours at the end.
One numbering everywhere on the site: Free intro call, 01 Discovery Phase, 02 Build, 03 Run. The details are below.
Where you are, what you want to achieve, and whether we are the right fit. If we are, you get a Discovery Phase proposal sized to your project.
1 hour · architectAn audit of what you have, or a plan for what you need: interviews, target architecture, requirements, timeline, estimate and risks. The result is yours.
Fixed · USD 5k–15kThe plan from Discovery, delivered with your engineers on your cloud account. A demo every two weeks, a weekly call, a monthly budget summary.
Fixed · T&M · retainerHandover with training and documentation. Your team runs it, or we stay on for maintenance and monitoring if that is the strategic choice.
Your team · or oursCommitments you can check in the contract.
Not a salesperson, not a consultant who has never built a system. One of our architects reads your brief, leads the intro call and sketches the architecture. The team takes over the day-to-day; the architect stays on through delivery.
Technology is a tool. Before we talk about how to build something, we ask whether it should be built at all: legally, clinically, commercially. If the business case doesn't close, we say so as early as we can, before you spend on a build.
When a project touches clinical workflows, our Chief Medical Officer is part of it from the start, not a reviewer bolted on at the end. Labeling protocols and what “good enough” means at the bedside get decided with a doctor in the room.
Code, models, cloud account, documentation: on your side, in writing. We build so your team can run the system without us.
Each stage has a clear boundary: what happens, what you get, and what belongs to the next stage. You can stop after any of them and keep everything produced so far.
It starts with a short brief, the form on our contact page, and an hour-long call with one of our architects, sometimes a follow-up. There is no script: the conversation follows what you're building and how far along you are. If you know what you want and why, we talk about how it could be built and what a Discovery Phase would cover. If you don't yet, we talk about what it would take to find out.
This is a conversation, not a quote: technical and business at once, and the moment to say whether we are the right team for it.
A clear next step: a Discovery Phase proposal sized to your project, a new system (A) or an audit (B), with its price, or an honest “not us”. No commitment either way.
The ROI analysis, the technical path, a review of your data and regulatory situation, a specification, estimates per module, audit findings or architecture design. The intro call ends at the fit; the Discovery Phase starts after it.
The Discovery Phase goes deep where the intro call stayed wide. We formalize the goals we talked about, then concentrate on the technology: how the system should be architected, what it must do, and what it must satisfy that no user ever sees – security, HIPAA/GDPR, uptime. Complex systems take longer than four weeks, and we'll say so before you sign.
Discovery comes in two shapes. A – a new system: we design it. B – an audit: you already have something, and you want to know what's sound, what's broken and what to fix first.
A specification and a roadmap for the software (A), or a diagnosis and a remediation plan (B). What you won't get: mockups. We map data flow and the engines behind the product, not screens – many systems we build have no interface at all. When one is needed, we build it or integrate with yours.
We build iteratively, against the milestones set in Discovery, and we build with you: your engineers are in the reviews from day one, so the result is co-created rather than delivered. How long it takes depends on scope – we'd rather tell you after Discovery than guess on a sales call. Current projects run from a few months to a year.
The commercial model is your choice: fixed scope when it's clear, time & materials with a budget cap when it isn't, or a monthly retainer for longer programs.
Working software on your own cloud account and in your repositories, accepted milestone by milestone – with validation on real data, a security review and documentation that grows with the code.
When the system is finished and its metrics validated, we hand it over properly: training, onboarding, documentation. You have full control – the system is designed to be operated by your team, so you never depend on us as a vendor.
Two paths from here. Your team runs it – the default, and the reason we document the way we do. Or, if running it isn't something you want to own for strategic reasons, we stay on for maintenance and support under a separate agreement.
Monitoring, retraining, audit support and the next statement of work – scoped and priced separately, never assumed.
Components proven across 25+ healthcare projects give every new one a running start: tested, documented, and adapted to your case rather than forced on it. You own the result.
Three people set the direction; a team specialized in the healthcare domain and in the technologies under it delivers. You'll meet the people who build your system on the first call and at every demo.
Former CTO with end-to-end technical ownership in three medical deep-tech startups, two of them later acquired (larger deal: USD 79M). At datarabbit he oversees the Discovery Phase and designs solution architectures with a focus on each project's impact and ROI.
Michał on LinkedIn
Radiologist with a PhD focused on AI in medicine; his research collaborations include leading labs worldwide, Stanford among them. Leads the clinical side of datarabbit's projects that involve clinical data or hospital workflows.
Assistant professor at Poznań University of Technology with a PhD in machine learning research. In the past he worked with supercomputing centers and drove technical work in EU R&D projects with combined budgets of over EUR 50M. At datarabbit he leads R&D, from model benchmarking to evaluation design.
Most of our clients are in the U.S. Here's how that works day to day.
25+
production data and AI projects delivered in healthcare and research
$79M
acquisition our diagnostic AI directly contributed to
60%
less documentation time in a live hospital deployment (clinical NLP)
40%
more experiments in the same time on a genomics MLOps platform
Nothing, and there's no catch, but there is a boundary. The intro call tells you whether we're the right fit and what a Discovery Phase would cover. The analysis itself, from ROI and technical path to specification and estimates, is the paid Discovery Phase.
A fixed price: USD 5k, 10k or 15k, depending on the scope – a standard analysis and a complex system are not the same job. If the number doesn't fit your budget, you'll know on the intro call, not after three.
Rarely. We build the engines – data flow, models, integrations – and many of our systems have no interface at all; they plug into yours. When a product needs one, we build it or integrate with your front-end team. We don't produce mockups in Discovery.
Yes, when scope is clear – usually after Discovery. Before that, time & materials with a cap, or a retainer. We'll recommend the model that fits the uncertainty in your project, not the one that's best for us.
Everything built for you is yours: code, models, documentation, to keep, change and run without us. Our reusable components, which make delivery faster, come with an unlimited license.
That's the default. We build on your account, in your repositories, with your team in the reviews – that's the co-creation we mean, and it's why they can run it after we step back.
We don't write the submission; your regulatory team does. We make sure the system can answer its questions: data lineage, training records, validation results, change control, and documentation a reviewer can read.
The intro call happens as soon as an architect has read your brief. The Discovery Phase usually begins within a couple of weeks after that. Build timelines come out of Discovery, not out of a sales call.
We'll say so as early as we can, often on the intro call, and suggest the cheaper thing: a data pipeline, a report, a rules engine. Several of our best client relationships started with “not yet.”
“The datarabbit team was brought in to help (a) bridge a staffing gap on our DevOps team, (b) accelerate a major cloud migration project, and (c) act as an expert consultant to pressure test our planned architecture. The team performed all three exceptionally well. I plan to engage them in future projects without hesitation.”
“What stood out was their ability to quickly research and answer questions at a level that was easy to understand – and their honesty about what AI could and couldn't handle.”
“datarabbit has a team of highly capable engineers that deliver with both quality and speed. The team provides trustworthy direction.”
“The datarabbit team was brought in to help (a) bridge a staffing gap on our DevOps team, (b) accelerate a major cloud migration project, and (c) act as an expert consultant to pressure test our planned architecture. The team performed all three exceptionally well. I plan to engage them in future projects without hesitation.”
“What stood out was their ability to quickly research and answer questions at a level that was easy to understand – and their honesty about what AI could and couldn't handle.”
“datarabbit has a team of highly capable engineers that deliver with both quality and speed. The team provides trustworthy direction.”
Book an intro call: an hour with an architect about where you are, whether we are the right fit, and what a Discovery Phase would cover. Honest, even if the answer is no.