Frans Moore

Frans Moore

I build AI systems that institutions can check.

In early 2025 I turned back to AI itself. Working with AI coding agents, since April 2026 inside a verification-gated pipeline I designed, I have built a control plane that runs a company with AI agents and treats every agent’s account of its own work as a claim, a live Ebola forecasting system that scores its misses in public, and open agent-trust infrastructure. Both ends of the problem matter to me: the data that goes into AI systems, and the evidence that what comes out of them is true.

I came to that work from the input side. After an MSc in International Public Health, I spent the years that followed on data collection, field research and health programs in eastern DR Congo, Tanzania and South Sudan, working on the lack of high-quality, organized data in niche, underserved contexts. A case count or a clinic register is data, and when it is wrong, the people it undercounts are usually the ones with the least room for error. Models are only as good as the data they learn from and the evaluations they are checked against, and data like this, from the places that need it most, is among the scarcest there is. That work gave me a multi-sectoral and humanitarian understanding that is rare in AI, and it is why I build AI for society’s direct benefit.

The question behind all of it started at the University of Georgia. From 2015 to 2019 I took courses through UGA’s Institute for Artificial Intelligence, including AI, discrete mathematics, cognitive psychology, sensation and perception, and mathematical modeling. I came away asking what the most consequential challenges are that AI, and thinking machines more broadly, could help us solve, and that question is part of why I went into public health.

Work

  1. Apr 2026 – present · Private system

    Earth: control plane for an AI-agent organization

    The system my company runs on. A fleet of AI agents does research, engineering and operations every day; an agent’s report of its own work is treated as a claim to be checked, not a fact.

  2. May 2026 – present · As of September 2026

    LOVS: outbreak evidence and forecasting system

    A live, cited evidence and forecasting system for the 2026 Bundibugyo Ebola outbreak. 87 forecasts were registered publicly before their outcomes were known and are scored in the open, misses included.

  3. Mar 2026 – present

    Agent trust infrastructure

    Open protocols and tools for verifiable agent identity and authorization: a paper, an open registry with outside contributors, and an SDK on npm.

  4. Jan 2022 – present

    Field research in eastern DR Congo

    Three years of mixed-methods human health and biodiversity research across conflict-affected South Kivu and Tanganyika provinces. The fisheries work is published as a peer-reviewed paper and an open dataset.

Writing

Articles

  1. Apolitical · September 2026

    AI Oversight and Verification Saturation

    Why oversight is the first institutional capacity AI saturates, and why it saturates silently.

  2. Apolitical · September 2026

    Hugging Face “Hack”: Six Questions Nobody Could Answer

    When agents can alter the logs they touch, a system’s own log is a claim, not evidence.

Papers

  1. Zenodo · 2026

    Composable Trust Infrastructure for the Agent Internet

    Six composable protocols for verifiable agent identity, capability discovery and authorization.

  2. Journal of Great Lakes Research · 2024

    Critical biodiversity, fisheries status and need for inshore fish communities conservation in Lake Tanganyika

    Peer-reviewed, co-authored. The biodiversity and fisheries status of Lake Tanganyika’s inshore fish communities, and the case for conserving them.

  3. MSc dissertation, Liverpool School of Tropical Medicine · 2020

    Adaptions by Immunization Programs in LMICs to an Infectious Disease Crisis: A Rapid Evidence Assessment

    How immunization programs in low- and middle-income countries adapted to keep services running through infectious disease crises. 5,806 records screened, 19 studies included.

  4. Honors thesis, University of Georgia · 2019

    An Exploratory Study of Job Satisfaction at an East African Medical Research Center

    A mixed-methods study of job satisfaction among 30 of the 36 staff at a Tanzanian medical research centre while it adopted Good Laboratory Practice.

Background

  • DR CongoField research, 2022 to present
  • South SudanProgram advisor, 2022 to present
  • TanzaniaResearch, 2017 to 2019
  • Liverpool School of Tropical MedicineMSc International Public Health, 2019 to 2021
  • University of GeorgiaBSc Psychology, Cognitive Science, 2015 to 2019

The full CV