FDA-VALIDATION ADVISORY
Validation paperwork is the bottleneck between your AI feature and your FDA submission. We use AI agents to compress the paperwork without compromising the rigor.
PROBLEM 1
Design controls, V&V, traceability matrices, risk files. Every release. Manual. Expensive. Slow. The team that ships the product is the team that documents it, and they do not scale.
PROBLEM 2
FDA's AI/ML guidance (Predetermined Change Control Plan, Good Machine Learning Practice, SaMD framework) is evolving. Off-the-shelf consultants do not yet know how to translate it into shipping code.
PROBLEM 3
Agents draft what your team would have drafted anyway. Your team reviews and signs. That is how AI gets into a regulated workflow safely, with accountability intact.
Three tiers. Fixed-price engagements with named deliverables. Discovery call is free.
TIER 1
gap analysis + 12-month roadmap
A senior outside read on your current AI/DSP/ML architecture in the context of FDA requirements, plus a 12-month prioritized roadmap.
TIER 2 · POST-REVIEW
scoped in writing up front
Scoped build of priority items from your Architecture Review roadmap. We deliver draft regulatory deliverables produced by our pipeline and reviewed by your qualified team, plus application code built under your design controls where in scope. You remain the manufacturer of record; we do not certify compliance and we do not deliver validated software.
TIER 3
rolling · 3-month minimum
Monthly engagement for clients who finished an Implementation and want ongoing access to the architect and the regulatory pipeline.
All engagements are a fixed fee agreed in writing before kickoff. No hourly billing, no surprise bills.
OUR APPROACH
AgentMedReg is our proprietary multi-agent architecture for FDA-regulated software development. Ten agents covering regulatory strategy, risk management, design controls, V&V, human factors, and CAPA, with a bookend pattern that opens and closes every engagement.
We invented it because we needed it. It runs on our own engagements before it runs on yours.
Read: When AI meets FDA →10 AGENTS · BOOKEND PATTERN
RIGHT FIT
NOT A FIT
FOUNDER BACKGROUND
Abel Villca, founder and architect. Background includes Roche and Silicon Valley health-tech startups. Has sat through pre-sub meetings, written design control documents, and shipped software and prediction ML algorithms under FDA regulation.
More about Abel and the team →BOARD OF ADVISORS REVIEW
Every engagement is reviewed by a matched advisor from our Board of Advisors. PhD in Computer Science with distributed-systems expertise (University of Geneva), PhD in laser physics and DSP (Belarusian State University), Daxtro SA embedded-systems industry experience (Switzerland). The matched advisor signs off on the architecture report's domain-specific section.
Meet the Board of Advisors →WHAT WE ARE NOT
TOOL ASSURANCE PACKAGE
Software used in a quality system has to be validated for its intended use by the manufacturer. Our drafting agents are that kind of software. With every engagement we hand your quality team what a risk-based assurance under FDA's Computer Software Assurance approach needs:
01
30 min call, free. Mutual NDA available. Decide if it's a fit.
02
2-4 weeks. Deliverable: gap analysis + roadmap.
03
If you continue: fixed-price build against the roadmap. 8-16 weeks.
04
Ongoing monthly engagement to keep design controls current.
No. We complement them. They handle submission strategy and FDA interactions; we handle the software, AI, and the agent-augmented documentation that connects the two.
No, and we do not claim they are. Two cases. Software we build into your product is device software: it is developed under your design controls and validated and released by you as the manufacturer. Our drafting agents are tools used in your quality system: you validate them for their intended use, and we supply the tool assurance package that makes that a risk-based, short exercise. In both cases every output is reviewed and approved by your qualified people. The agents do not sign off; your team does.
Part 11 considerations are part of every engagement. We help you map workflows to Part 11 requirements; we do not sell Part 11 software as a product.
We actively track FDA's AI/ML lifecycle program, GMLP, PCCP guidance, and SaMD framework. Public references on our blog.
Yes, mutual NDAs as standard. Template available on request.
Yes. Our deliverables are designed to slot into your QMS, not replace it.
Seattle, WA. Architecture, sales and client-facing work are Seattle-based, with full overlap with US business hours. All client-facing work is in English.
30-minute discovery call. Free. Mutual NDA available. We use it to decide whether we are a fit before anyone signs anything.
Book a discovery call