AI strategy and working systems for health-tech

Find the AI use cases that matter for your business, then get them built. I design and ship agentic workflows for regulatory and commercial operations, and I have the shipped platforms to prove it.

Let's Talk
Why This Practice Exists

Every health-tech board is asking about AI. Most companies under 200 people respond the same way: a few ChatGPT licenses and a stalled pilot, followed by a six-figure quote from a dev shop that has never read an FDA guidance document.

The hiring market explains why. Postings for hybrid "strategy plus build" AI roles grew several-fold over the past year, and a full-time head of AI in healthcare runs $250K to $350K in base salary before equity. Companies post these roles, the roles sit open for months, and the work doesn't get done.

The talent pool splits cleanly. AI consultants who have never shipped a production system. Engineers who have never worked under HIPAA or sold into a health system. People who can do both, in healthcare specifically, are rare. That gap is what this practice fills.

What I Build

Fixed scope, fixed timeline, working software at the end. Recent and representative builds:

BD intelligence systems

Agents that find and qualify companies against your ICP, then draft outreach for human review. One of these runs inside a clinical trials consultancy today, scanning trial registries for biopharma sponsors with imaging-heavy pipelines.

Market and competitive intelligence engines

Standing workflows that track competitor moves and FDA clearances in your segment, then deliver a structured brief on your schedule instead of an analyst's.

Regulatory research workflows

Agentic research over FDA databases: predicate searches and safety signal review. The 510(k) Predicate Finder at keenr.ai is a working example you can try right now.

Founder content engines

A system that turns your expertise into a steady publishing pipeline: source research and drafts in your voice, inside an editorial workflow you approve. You stay the author. The system removes the blank page.

If your bottleneck isn't on this list, the assessment below exists to find it.

How Engagements Run

Most clients start with an AI opportunity assessment: two to four weeks mapping your workflows and ranking use cases by impact and feasibility. You get a build roadmap with a build-versus-buy call on each candidate, whether or not we work together afterward.

From there, builds run as fixed-scope sprints measured in weeks. Each one ends with working software and someone on your team trained to run it. Companies that want ongoing leadership keep me on as a fractional AI lead, typically one to three days a week, to own the roadmap and keep shipping.

Engagements start with a conversation about what's actually slowing you down.

Shipped Work

Keenr AI

I founded Keenr AI, an agentic platform for medtech regulatory research, now coming out of stealth. Multi-agent workflows over FDA data that turn weeks of regulatory analysis into hours.

510(k) Predicate Finder

A live tool at keenr.ai. It classifies a device description, searches predicates across FDA databases, runs safety checks against MAUDE and recall data, and produces an analysis report. Try it.

BD agent system

A three-agent workflow built for a clinical trials imaging consultancy: one agent finds biopharma leads with imaging-heavy trials, one qualifies against the firm's ICP, one drafts outreach. A person reviews everything before it sends.

Eighteen years in the domain

PhD in biomedical engineering with a focus on medical imaging, plus an MS in machine learning. Eighteen years commercializing healthcare technology across consulting and medtech SaaS. The systems work because the domain knowledge underneath them is real.

Who This Is For

This work fits health-tech and medtech companies between roughly 10 and 200 people that need AI capability without a full-time hire. It also fits healthcare consultancies and CROs that want to turn their expertise into AI-backed products. Timing matters too: a 510(k) under review, a US market entry, or a fundraise are all moments where these systems pay for themselves fastest.

Questions That Come Up
Can't we just use ChatGPT?
For drafting emails, yes. The systems I build are different in kind: multi-agent workflows connected to your data sources, with review gates and evaluation checks built in, running on a schedule whether or not anyone remembers to prompt them. A chat window is a tool. These are operations.
We're in a regulated space. Can we even use this?
That constraint is the reason to work with someone who knows the domain. Builds account for HIPAA from the architecture stage and keep humans in the loop where risk demands it. Anything touching FDA-regulated device functions gets scoped explicitly with your regulatory counsel. I have spent 18 years inside these constraints.
Build or buy?
Sometimes buy. The assessment ends in a build-versus-buy call for each use case, and I have no incentive to push builds you don't need. Maintenance retainers only make sense for systems that earn their keep.
What happens when you leave?
You keep the working systems and the documentation, and someone on your team is trained to run them. Handoff is part of every scope. The goal of fractional work is to make your team capable, then get out of the way.
Why not hire a full-time head of AI?
If you can find one with healthcare depth and the budget for a $300K+ package, you should consider it. Most companies at this stage can't, and the role sits open for months while the work waits. Fractional gets the roadmap and the first systems shipped now, and makes the eventual full-time hire easier to scope.

Start with a conversation

Tell me what's slowing your team down. If there's an obvious AI answer, I'll tell you. If there isn't, I'll tell you that too.

Let's Talk