Independent research · Agentic AI · Regulated industries
MAX Research Collective researches and builds AI for healthcare and other regulated industries: run where its owner has control, keeping the data in, and leaving every decision that matters to a person. What we learn building it is published here.
Search the library
Start with a topic
Regulated, audited & security-first
Local-first by default — MAX3, REEF, and FeedHacker keep data on your own hardware, so there's no third-party egress of PII/PHI, the smallest possible breach surface, and a clean data-residency story. Automation stays approval-gated, reversible, and audited.
The other half of “where the rules are real” is the half that lands in the financial statements. Which engineering hours capitalise and which expense; what a useful life does to a depreciation schedule; what §174A changed about the cost of outsourcing; whether an AI budget is one line or three. These are audited too, and the evidence is your ticket history. Career & leadership writing →
This site: HTTPS only · no third-party trackers · no analytics cookies · security.txt
Track record
Latest
Agents write a slice in minutes. The slow part was always deciding whether it works and whether anyone wanted it, and that part is still yours.
HAL 2000 learned to know who was speaking, MAX3 to grade every claim, MAX4 to own every seam. MARA inherits all three before she meets a patient.
Every skill MAX4 learns starts switched off and earns the right to act, one held-out test at a time. That ladder is the bar MARA will be held to.
Build log
Modular Algorithmic eXecutor (MAX): a local-first, voice-first AI resident built as a single Python process — a DuckDB graph spine for memory, a Forward-Forward learner that predicts, proves and adjusts, and a voice loop to talk to MAX. The collective's flagship.
Medical Agentic Resident Assistant (MARA): MAX pointed at healthcare. One ankle injury touches fifteen people in a dozen systems; MARA carries the record, coverage and history across every one of them, recommends to each person in the chain, and learns from what actually happened afterwards. MARA never signs: a person makes every decision.
A local-first network-visibility and gateway-defense tool — a named fleet of sensors fused into one high-confidence card that sees it, explains it in plain language, and acts only on your approval.
Founder
Jason Newell leads organizations and people in technology, and is a researcher, a deep technologist and a builder. He started as a software developer on a CRAY and has gone where the technology is ever since — networks, data, cloud, and now AI. He founded MAX Research Collective to build agentic AI for regulated industries that runs where its owner has control, with healthcare as its strongest focus: MARA, on the MAX3 core.
New research, projects and field notes as they publish. Following happens on LinkedIn — we never see an address, and there is nothing to unsubscribe from.
An occasional summary of new research, each with a link to the full piece.
Confirmed opt-in: we send one link to check the address is yours, and you are not on the list until you open it. Every email carries a one-click unsubscribe that removes you from the newsletter only — never from LinkedIn.