Health systems.
NeoCare Labs is a senior studio for AI and agentic systems in healthcare. We take health products from idea to production: agents that do real work, full-stack engineering, and cloud infrastructure.
Experience from teams at
The stack we ship with
One studio, end to end
From the first model experiment to the app in your users' hands: the same senior engineers own every layer.
AI & LLM engineering
Agents, RAG, and LLM features for the clinical context: visit summaries, documentation support, and assistants for care teams, with evaluation, guardrails, and cost control from day one.
Full-stack product engineering
Patient portals, clinic systems, and health apps built end to end in JavaScript and TypeScript: Next.js and React on the web, React Native on mobile, Node.js and Python services behind them.
Cloud, data & MLOps
AWS, Google Cloud, and Azure infrastructure designed for sensitive health data: serverless APIs, CI/CD, monitoring, and observability, with privacy and compliance treated as project requirements.
Fractional AI leadership
A Lead AI Engineer on the side of clinics, healthtechs, and providers: AI strategy and roadmaps, architecture reviews, hiring support, and hands-on enablement for your team.
Agents that do real work
Beyond chatbots: we design and ship agentic systems that plan, use tools, and complete whole clinical workflows, from patient intake to documentation, with the reliability of production software.
Multi-agent orchestration
Planner and worker architectures for clinical and admin workflows: intake, scheduling, documentation, and billing coordinated on LangGraph, CrewAI, and AutoGen.
Tool use & integrations
Agents that safely operate your clinic's systems: records, calendars, billing, and internal tools, connected through MCP and function calling with strict permission boundaries.
RAG & knowledge
Retrieval pipelines grounded in protocols, clinical guidelines, and your operation's data, so answers cite their source instead of inventing it.
Evals, guardrails & observability
In healthcare, mistakes are expensive: every agent ships with an evaluation harness, tracing, cost budgets, and human-in-the-loop checkpoints on sensitive decisions.
Products we've built
We build health products for other companies and for ourselves. HealCare and Falo are our own, taken from first sketch to production in the studio, the same way we build for clients.
Falo
AI-powered accessibility app
Our flagship case: an AI-powered augmentative and alternative communication (AAC) app for nonspeaking people, built from scratch by the studio and in open beta today. It is our slogan in practice: we took a high-friction corner of health, assistive communication, and modernized it with real AI.
- A pictogram board built in seconds from a photo, voice, or a few taps
- Automatic Portuguese conjugation, a first among major AAC apps
- On-device speech that works offline
- React Native on mobile, Python backend, vision and LLMs behind it
HealCare
Clinic platform with an AI assistant
A healthcare management platform for therapy clinics: patients, scheduling, medical reports, and documents, with role-based access and an AI clinical assistant woven into the daily workflow. Next.js frontend, Python API, deployed on Google Cloud.
How an engagement runs
Short cycles, working software early, no black boxes. You see progress every week.
Discover
A focused week to nail the problem: goals, users, data, feasibility, and risks. You get a concrete plan with scope, timeline, and cost.
Prototype
A working prototype of the riskiest part first, usually the AI core, in weeks, not months. We validate with real data before committing to the full build.
Build
Production engineering in weekly increments: clean architecture, tests, security, and a staging environment you can click through from day one.
Ship & scale
Launch on solid cloud infrastructure with monitoring and CI/CD, then iterate on real usage. We stay on for evolution or hand over with full documentation.
Rômulo Amaral
Founder & Lead AI Engineer
Senior engineering, no hand-offs
Lead AI Engineer with a decade of experience shipping AI in environments of very high scale and technical demand, today focused on agentic systems: agents that plan, use tools, and complete real work in production. Engineering background from UFRJ and the University of Waterloo.
He built his career across countries and industries, from Canada to Ireland. At PokerStars, he worked on systems that processed millions of simultaneous events in real time; later, as a Lead AI Engineer at Johnson & Johnson, he applied artificial intelligence to one of the world's most critical sectors: healthcare. He is also married to a speech therapist, seeing the daily reality of clinical work up close, and that is where his fascination with the field comes from.
At NeoCare Labs, that experience turns into product: every agent is designed, evaluated, and deployed under his direct lead, with the conviction that well-built agents remove busywork and put the focus back on what matters to your users.
The person you talk to does the work
No account managers, no juniors learning on your budget. Direct communication with the engineer building your product.
Production-first mindset
Evaluation, monitoring, costs, and failure modes are part of the design from the first sketch, not an afterthought.
Remote-first, timezone-friendly
Years of experience working with distributed teams and clients across time zones, with clear async communication.
Common questions
What kinds of projects do you take on?
AI-powered health products end to end (web and mobile), LLM features added to existing platforms, agents for clinical and admin workflows, and cloud/MLOps work. If it involves putting AI in front of care teams and patients, it's a fit.
How do engagements work?
Every project is priced for the entire scope of work, based on its complexity. We agree on scope, deliverables, and a single fixed price before we start. No monthly billing, no hourly meters, no surprises.
Who actually does the work?
Senior engineers only, led directly by the founder. We keep the team deliberately small and only take on a new project when we have the capacity to give it our full attention. If we are at capacity, we say so and offer you the next available slot.
Can you work with our existing team?
Yes. Alongside full builds, we embed with in-house teams for AI architecture, code reviews, and enablement, leaving your team stronger than we found it.
Who owns the code and IP?
You do, fully. We sign NDAs before hearing anything sensitive, and every line of code, model, and document produced during the engagement belongs to you.
How do we get started?
Send a short note about your product or problem. We reply within two business days with honest first thoughts and, if it's a fit, a proposal for a discovery week.