Bauer and Yonclas spent six years building AI infrastructure for the Army and the intelligence community before a single enterprise customer ever saw it. The architecture they built inside T2S Solutions, a defense and intelligence contractor backed by Madison Dearborn Partners, spun out in December as Axonis, and named Todd Barr chief executive to take it to market.
Barr's résumé runs through Red Hat, Chainlink, and GitLab, three companies with nothing to do with defense contracting. His job is to translate what David Bauer, Chief Technology Officer and technical co-founder, spent two years incubating inside a government contractor into something a hospital system or a manufacturer can buy.
The pitch comes down to one architectural choice.
AI Goes to the Data Instead of the Other Way Around
Axonis places a node wherever a customer's data already lives, then runs models against it in place, without copying that data into a shared environment first. The company describes four capability layers a customer can run this way: direct AI integration, entity matching across sources, scoring and probability models, and traditional machine learning prediction (Axonis, 2025).
Above those four sits what the company calls a decision intelligence layer. Every time a model surfaces a recommendation, the system logs the data snapshot, the model output, and a written human justification for whatever the person did next, act or not act. Axonis built that attestation trail for a defense customer base where every action has to be auditable after the fact. Axonis is now positioning the same capability for healthcare, financial services, insurance, and critical infrastructure, industries that carry a similar audit burden without the military label attached.
"The real barrier to AI isn't modeling, it's getting AI models into production."
Who Buys Infrastructure Like This
In a conversation at Axonis's booth, Barr described the buyer as whoever owns a business outcome sitting on top of data that has stayed scattered for reasons nobody plans to fix. A hospital system with records that never merge across facilities. A manufacturer running equipment old enough that pulling data off it costs more than the data is worth. Barr calls this decentralized-by-design data, distinct from data that stays scattered because a company has not gotten around to consolidating it.
Axonis positions itself as complementary to platforms like Snowflake and Databricks rather than in competition with them, running AI on the data that never made it into those centralized environments in the first place.
A Human Still Signs Off on the Decision
Barr told the author that most AI infrastructure pitches lead with speed. Axonis leads with something else: a person still reviews the decisions the platform is built to support. An operator does not need a millisecond response to know a machine is drifting out of tolerance. A clinician does not need real-time inference to review a flagged chart before a shift ends. The system generates the score and the audit trail, and a person decides on a timeline the business already runs on.
Axonis. "Axonis Launches Federated AI Platform, Names Todd Barr CEO." Axonis, 10 Dec. 2025, axonis.ai.
Barr, Todd. Conversation with the author. Axonis booth, Ai4 2026, The Venetian, Las Vegas, 4 Aug. 2026.
