Talon AI · Shreveport, Louisiana

We build the engines
healthcare decisions run on.

Concinna — the auditable answer engine — is the flagship: decision-ready intelligence from billions of rows of governed public record, every number arriving with its receipts. Around it, the compliance engines and decision systems that put those answers to work. All of it in production, for paying clients, today.

Billions
of governed rows
9.5M
providers on file
A dozen+
named public authorities
Zero
PHI in the corpus

Touch any number for its receipt — that's the house rule.

The flagship

Intelligence, arranged with grace.

American healthcare publishes an extraordinary public record — payments, prescribing, coverage rules, prices, ownership, conduct — scattered across thousands of files that were never designed to be read together. Concinna resolves that record into one coherent picture: every provider, organization, drug, procedure, and dollar joined on a common spine and kept current on the source's own schedule.

On top of that corpus, Concinna generates answers — market reads, target dossiers, risk and rate analyses — packaged for the decision they serve, not the database they came from.

Drawn from the named public record:
CMS · FDA · NPPES · Open Payments · HRSA · OIG · SEC · IRS · DOL · NLM · State boards & registries

Every answer carries its receipts. Sources are named, vintages are stated, and a modeled figure is labeled as modeled — never passed off as observed. We would rather show our work than ask for trust.

The Concinna principle
Visit concinna.ai →

The system above the model

Models are replaceable.
The reasoning system isn't.

We don't train foundation models, and we don't bet the company on any one of them. What we built sits above that layer: the entity graph that knows who's who in American healthcare, the identity resolution that joins a fractured record into one, and the trust logic and provenance that make every answer auditable.

That system is difficult and expensive to reproduce — and it compounds. Every source onboarded makes the graph harder to match. Every verified answer makes the next one cheaper. Swap the model underneath tomorrow; the asset doesn't notice.

Proving ground

Tested where the margin is thinnest: value-based care.

Risk-bearing organizations live on a few basis points of medical margin. The questions that decide their year — which market to enter, which partners fit, where referrals leak, what the risk pool really looks like — demand answers that survive an actuary. That is the standard the engine is built to.

In production

Market-selection, risk, and network analytics for a national value-based-care enabler — county-level market reads, partner-fit dossiers, and risk-model outputs delivered as working product.

In production

Pre-submission documentation auditing for a multi-provider transitional-care organization — clinical documentation checked against live coverage rules before a claim ever leaves the building.

Client names are shared in conversation, under the confidentiality their engagements deserve.

What we build

A house of engines. One discipline.

01

Concinna

The answer engine itself — the governed public-record corpus, the entity spine that joins it, and the analytics generated on top. The flagship, and the foundation under everything else we ship.

concinna.ai →
02

Compliance & chart-audit engine

Documentation checked against live coverage rules — LCDs, NCDs, payer policy — before submission, where a denial costs nothing to prevent. The same governed rule corpus behind Concinna, pointed at compliance. Running daily.

In production
03

Decision packs & bespoke intelligence

Entity-resolved dossiers, market reads, rate and negotiation analyses, diligence support — scoped engagements built on the same corpus, with the same receipts.

Start a conversation →

How we work

Discipline is the product.

Verify first

No number ships until it passes an accuracy gate against the live corpus. Findings are confirmed on real source data — not on the output of the tool that produced them.

Provenance, always

Every figure names its public authority and its vintage. Where a value is modeled or estimated, it says so on its face. An answer you cannot trace is an answer you cannot use.

Public record only

The corpus is built entirely from governed public records — no patient-level data, no PHI. Client work that involves protected health information runs separately, under executed BAAs, on covered infrastructure.

Fresh on the source's schedule

Each dataset refreshes on its authority's own publication cadence, with automated freshness checks. Data that has quietly gone stale is treated as an outage, not an inconvenience.

The company

"Built from operational reality, in production before it was ever a pitch."

Talon AI was founded by Matt Rimmer in Shreveport, Louisiana. The company is privately held and product-led: every capability described on this page is running in production for paying clients, built against real operational pain rather than theory.

The stack is ours end to end — models, pipelines, and the data corpus run on infrastructure we own. Nobody meters our questions, and nobody else holds the keys.

We keep the work honest the same way we keep the data honest — claims you can check, engagements that start smaller than the ambition, and a preference for showing rather than telling.

Contact

Tell us what you're trying to answer.

A market you're weighing, a partner you're vetting, a denial pattern you can't see the bottom of — start with the question. We'll tell you plainly whether the engine can answer it.

We take a small number of engagements at a time. Serious inquiries move quickly.

matt@runtalon.ai
318-401-2221
Book a call directly ↗

By submitting this form you agree to our Privacy Policy.