Steve Moses

Steve Moses

I run Co-Intelligence Labs, an AI product and advisory firm. Before this, I built healthcare at Amazon, researched AI at Cambridge, and sold tech at Microsoft.

Get in touchMy background

Background

I take AI from research to product to go-to-market.

Outcomes

The vector sum of me

At Co-Intelligence Labs, we build AI products and advise leaders on AI strategy — from agentic AI supporting thousands of seniors, to engagements with C-suite executives at Fortune 50 companies, family offices, and startups. Occasionally I give talks — like at JPMorgan Chase, where I was rated the top speaker at their Innovation Week.

Before that, I grew Amazon Pharmacy's enterprise business from the ground up to multi-nine-figure annual revenue in three years. Earlier, I spent four years at Microsoft leading enterprise sales and partnerships for global publishers — The New York Times, News Corp, Time, Hearst, Bloomberg — delivering 200%+ quota attainment on a nine-figure portfolio, and helping turn a machine learning research paper into one of Microsoft's first ML products, Azure Personalizer.

In between Microsoft and Amazon, I stepped out of industry to train as a researcher at the University of Cambridge, studying personalized AI for behavioral health. Earlier still, at Accenture, I advised the Obama Administration on healthcare and Tanzania's Minister of Education on national education strategy.

I fell in love with the potential of AI somewhere in the middle of all that, and I've been at it ever since.

Organizations I've worked with

  • Amazon
  • Microsoft
  • JP Morgan Chase
  • University of Cambridge
  • Accenture
  • The New York Times
  • News Corp
  • Time
  • Hearst
  • Bloomberg
  • American Express
  • The Wall Street Journal
  • UCLA
  • WebMD
  • Prime Therapeutics
  • Blue Cross Blue Shield
  • Pioneer Square
  • 25 Madison
  • The Neuroleadership Institute
  • Biohaven

Research

AI that adapts to the person

My research has been focused on AI personalization — how a system can adapt to the way someone thinks, communicates, and decides. Primary applications have been behavioral health and remote care companions.

Academic positions

Visiting Researcher
University of Cambridge Behavioral Health Lab
Reviewer
The New England Journal of Medicine Catalyst (AI/ML applications)

Education

Cambridge
MRes · PhD candidate (on leave)
Duke University
MBA (Klopman Scholar)

AI measurement

  • The Grounding Series: six controlled studies of how books shift what language models say

    Does grounding change model answers, or is it incantation? Registered designs with measured noise floors and fictitious-book controls, judged blind across model families — thousands of generations on Claude and Grok.

    2026 · reports, code, and data

  • Which Model? Value-profiling five AI assistants with psychometric instruments

    A probe battery adapted from MFQ-2 and Schwartz values locates ChatGPT, Claude, Gemini, Grok, and Kimi on seven value dimensions with bootstrap CIs — plus a benchmark of how well each serves older adults.

    2026 · in progress

Behavioral health

  • Improving Weight Loss Adherence: A Machine Learning Approach to Personalized Chatbot Interventions

  • Using Qualitative Chatbot Messaging Content to Understand Anxiety and Hope Causation During Covid-19

  • Using Text Messaging Chatbots to Measure Well-being During Covid-19

  • Healthy Food Delivery Combined with SMS Text Follow-up Improves Healthy Eating