Our record

Our record, stated plainly.

Artheus was formed in July 2026. While our past performance grows, here is our current engagement and the work we have built and measured ourselves.

Built and measured

What we can show you today.

Four pieces of our own work, none of it for clients: the kind of work we offer companies and agencies, done on ourselves first.

0.974
F1 score recovering cable connectors (the cable runs between equipment) from a synthetic test corpus
1.000
Provenance: every extracted element traced to its source page, same corpus
9
White papers published on applied AI, security and modernization
110/110
NIST SP 800-171 Rev. 2 self-assessment score, posted in SPRS in September 2026
The detail

What each result means.

Each number comes with its limits.

01

Current consulting engagement

In progress since July 2026

Since July 2026, Artheus has supported a federal services contractor under a consulting agreement, with advisory and technical work for its home office.

We do not name clients without their permission.

02

Document-to-model pipeline

Technology readiness level 3

Our pipeline turns mixed-format engineering documents into a populated engineering model, with every element traced to the document, page and row it came from. We built and measured it at our own expense, on a synthetic test corpus with no controlled or classified material.

An engineer on a program or at a plant can go from any element straight to its source page. The planted defects test whether disagreements between documents are flagged instead of silently merged.

Test results on a synthetic corpus
18documents in the synthetic corpus
7different file formats
40defects planted on purpose
0.974F1 score recovering cable connectors (the cable runs between equipment)
1.000F1, requirements recovered
1.000provenance resolvable for every element
37.5sone timed run of the full corpus on a single CPU workstation
TRL 3technology readiness level: proof of concept

The corpus is synthetic and we built it, so these numbers show the method works; they do not predict how it will score on your documents. On an engagement, we measure on a sample of yours before agreeing a target. Cable connectors and requirements are our two strongest element types; others score lower, and the full scorecard is available on request.

F1 is a standard accuracy score from 0 to 1 that counts both missed items and false finds. A score of 1.000 is perfect.

See Digital Engineering
03

FirstShift

Working prototype

FirstShift is an Artheus product in development: adaptive, multilingual AI that checks that frontline workers understand safety-critical procedures, in their own language.

Today it is a working prototype. It is not yet a fielded product.

MultilingualFrontline safetyAdaptive learning
See Training & Readiness
04

Nine published white papers

Published

Short, practical papers on the problems we work on, from running AI inside a security boundary to testing it before anyone depends on it.

Applied AIZero TrustModernizationAI governanceData readinessWorkforce
Read the white papers
05

Our own security posture

Affirmed in SPRS

Artheus went from formation in July 2026 to a NIST SP 800-171 self-assessment score of 110 of 110 and CMMC Level 2 (Self) status in about two months. The score was posted and the status affirmed in SPRS (the Defense Department's Supplier Performance Risk System) in September 2026.

A self-assessment is our own scored review against the standard, not a third-party certification. Our CMMC and NIST SP 800-171 Readiness service follows the same path.

NIST SP 800-171CMMC Level 2 (Self)SPRS
See CMMC readiness
Start here

See whether we are the right fit.

Tell us what you are working on, and we will tell you honestly whether we are the right fit. Registration details and our capability statement are on the contracting page.