Kaermorhen
Starkville, Mississippi

The platform knows first. Kaermorhen is building what listens.

Kaermorhen builds onboard prognostics for military ground fleets. The system rides on the platform, monitors the condition of its major components, and warns the maintainer before a part fails, while there is still time to requisition it and schedule the work. It runs with no reachback, meaning no satellite or network link to a rear headquarters. And it stays advisory. The maintainer retains the decision.

Service the part on condition, not on the calendar.

Kaermorhen works inside a sustainment strategy the Department has already established and directed. The terminology is technical. The logic underneath it is not.

Condition Based Maintenance Plus

Legacy sustainment runs on two models. Scheduled maintenance pulls a component at a fixed interval whether or not it needed pulling. Corrective maintenance waits for the failure and repairs it after the fact. The first discards serviceable life. The second deadlines the vehicle on a timeline the unit did not choose, and deadlined is the Army's word for a platform that cannot perform its mission.

Condition Based Maintenance sets the interval by the measured condition of the component instead. Sensors observe the part while the platform operates, and the work order follows the evidence. The Department's version is Condition Based Maintenance Plus. DoD Instruction 4151.22 established it as the primary sustainment strategy, directed for new systems and for integration into legacy fleets.

Prognostics, or Predictive Maintenance

Condition tells you the state of a component today. Prognostics estimates its remaining useful life and the probable point of failure. That forecast is what converts an unscheduled failure into a scheduled induction.

The difference shows up in readiness. With warning, the unit requisitions the Class IX repair part, which is the supply class covering components, positions it forward, and puts the vehicle in the shop on a day it can afford to lose one. Without warning, that same component lets go during a movement, the platform is deadlined in place, and recovery multiplies the cost. Prognostics is what the policy asks for. On ground platforms it is the piece still largely missing.

Sustainment cost is climbing while readiness falls.

Roughly 70 percent of what a weapon system costs across its life is operations and support, not the price of buying it. When a component fails without warning the platform goes non mission capable, the repair runs higher than a planned one would have, and the depot backlog lengthens behind it. The indicators are usually present in the data before the failure occurs. On most ground platforms, nothing is watching for them at the component level.

$2.3B Army and Marine Corps combined spend on ground vehicle depot maintenance, FY2023.
0 of 18 Vehicle types reviewed that met the Army's 90 percent mission capable goal in FY2024.
70% Approximate share of total life cycle cost consumed by operations and support.

Source: GAO 25 108679, Weapon System Sustainment, September 2025.

Warning at the component, ahead of the failure.

Components signal degradation before they quit. Vibration signatures shift. Thermal profiles climb. Acoustic and electrical behavior drifts off baseline. A seasoned maintainer catches some of this during PMCS, the preventive checks a crew runs before and after operation. The Kaermorhen system monitors across every operating hour and reports when a component departs its normal envelope.

Onboard

Operates without reachback

Processing happens on the platform. In a denied, degraded, or intermittent communications environment, the capability does not degrade with the link. Contested logistics assumes the network is itself a target.

Forecast

Time to source the part

Warning turns an unscheduled deadline into a scheduled induction. The unit requisitions ahead, positions the part, and takes the vehicle down on its own timeline instead of the failure's.

Advisory

The maintainer retains authority

The system reports what it observed and what it recommends. It issues no directives. It is built to be overruled, and overruling it costs the maintainer nothing.

What is not on this page. System architecture, sensing methodology, model design, and unit economics are held close. Kaermorhen discloses them under NDA to investors, program offices, and prime partners. Anyone for whom that detail is decision relevant should request it directly.

Built for the military. Useful well beyond it.

Kaermorhen is actively pursuing contracts across American manufacturing, automotive, and agricultural industries. The failure that deadlines a tactical vehicle also stops a production line, sidelines a fleet truck, or strands a combine in the middle of harvest, and the economics of catching it early run much the same way in each case.

Manufacturing

Lines that cannot stop

Unplanned downtime on a production line carries the same shape of cost as a deadlined platform. The equipment is different. The failure physics and the value of early warning are not.

Automotive

Fleets under load

Commercial and municipal fleets run hard, on schedules with no slack, and absorb the same penalty when a component lets go without notice on the road.

Agricultural

Windows that do not reopen

Harvest is measured in days. Equipment that fails inside that window costs a season, which makes early warning worth more in agriculture than almost anywhere else.

The target customer is the United States military. Everything Kaermorhen builds is held to the standard that customer sets. Commercial work broadens the installed base and hardens the product without pulling the company off the mission that defines it.

Lines that do not move.

American built

The company is American. The founders are U.S. citizens. Development, integration, and support are performed in the United States, and the intellectual property stays in American hands.

The maintainer retains the decision

Automation bias is a documented failure mode, not a theoretical one. A crew that has handed its judgment to a display becomes a vulnerability, and a capable adversary will eventually find it. Overreliance on AI will cost the force more than it returns.

If a Kaermorhen system is ever the reason a soldier stops thinking, the wrong capability was fielded.

Readiness first

Kaermorhen measures itself against a single question. Are more vehicles mission capable. Not features delivered, not data accumulated. If a capability does not move the readiness rate, it has not earned its weight, its power draw, or the maintainer's attention.

Data ownership

The raw maintenance data a platform generates belongs to the United States military. Kaermorhen holds no ownership interest in it and no authority to sell it, share it, or transfer it to an outside party. It stays inside the military specific Controlled Unclassified Information environment it was collected in, and it is never used for any purpose outside that environment.

The models trained on that data are proprietary to Kaermorhen. The record belongs to the government. The engineering built on top of it belongs to the company. Both hold at once, and neither weakens the other.

Handled to the government standard

The Department already wrote this standard, so Kaermorhen conforms to it rather than improvising around it. Program data is marked and handled under the DoD Controlled Unclassified Information program. Any Kaermorhen system that stores, processes, or transmits it is built against the 110 security requirements in NIST SP 800 171, the baseline DFARS 252.204 7012 imposes on a contractor holding covered defense information.

Where it resides

Most of it never transits anywhere. Processing occurs on the platform, so the data stays with the machine unless the government directs otherwise. Less movement is less attack surface.

Where storage off the platform is required, it resides in a government cloud region authorized at FedRAMP Moderate or above, not commercial cloud. Program data does not touch commercial email, consumer file sharing, or this website.

Two seats filled. Three open.

All three open seats are cofounder seats carrying equity from the start, not hires. Anyone whose discipline matches one of them should write to the address below and describe what they have built and what of it has been fielded. Every seat requires U.S. citizenship.

Founder

Eli Herrington

PhD Student of Mechanical Engineering, Mississippi State University

  • Mechanical Engineering
  • Research in Economic Development of Defense Industrial Base
  • U.S. Citizen

Cofounder

William Mayfield

Mathematician and Mechanical Engineer

  • Mathematics
  • Mechanical Engineering
  • Modeling and Analysis
  • U.S. Citizen

Cofounder seat open

Software Engineering

Embedded and systems software

Software that runs on constrained hardware, in the field, with no network to fall back on. Real time and resource limited experience carries far more weight here than a long list of frameworks.

Cofounder seat open

Security Engineering

CUI posture and product security

Ownership of NIST SP 800 171 implementation, the system security plan, SPRS reporting, and security designed into the product rather than appended to it before an assessment.

Cofounder seat open

Finance and Accounting

Books, contract accounting, and the SBIR path

Ownership of the books, government contract accounting, records that withstand a Defense Contract Audit Agency review, and the financial path through SBIR Phase I and Phase II.

Get in touch

Kaermorhen is raising its first round.

Kaermorhen is in conversation with investors, program offices, and sustainment partners who want ground fleet readiness solved at the platform.

contact@kaermorhenproject.com
  • Starkville, Mississippi
  • Stage: Pre seed
  • SBIR Phase I in work