Aerospace Runs Two Certification Clocks, and Only One Is Keeping Time

Two certification clocks, one aerospace industry

Three MRO stories landed in the same week with the same quiet message: conventional aerospace certification is running at full speed. Prima Air picked up FAA Part 145 maintenance certification for its Hollywood facility. Blackhawk Group added anti-icing and propeller STCs and launched mobile support out of Colorado. Stevens Aerospace hit a milestone on the 1,000th Garmin G1000 NXi retrofit for the Army’s C-12 fleet. Wheels Up signed Av8 to a five-year landing gear support agreement ahead of fleet expansion. Each of these is a Part 145 or STC transaction, a well-worn pathway that regulators, shops, and operators all know how to price, schedule, and audit.

Set that against a different signal from the same week. A whitepaper flagged in Military Embedded Systems’ edge AI coverage is titled, plainly, “Efficient verification through the DO-178C life cycle,” applied to drone perception systems. The fact that this needs writing at all is the tell. DO-178C’s life-cycle model assumes deterministic, traceable software behavior that a reviewer can map requirement to code to test case. Machine-learned perception models, the kind now embedded in edge AI drone payloads, do not decompose that way. There is no equivalent of an STC for a perception model. There is no Part 145 shop that signs off on a neural network’s drift the way Blackhawk signs off on an anti-ice system.

Why this matters for the buyer

For a procurement or airworthiness lead evaluating an “AI-enabled” avionics upgrade, the operative question is not whether the vendor has a certification. It is which certification, and what it actually covers. An STC for a sensor housing or a Part 145 approval for an install shop says nothing about the assurance status of the model running inside that sensor. Buyers who conflate the two are importing conventional certification confidence into a domain where the underlying verification standard has not caught up, a gap regulators have not closed and vendors have little incentive to flag.

The practical move is to unbundle every AI-adjacent retrofit quote into its component certifications: what’s covered by existing STC or Part 145 scope, and what’s riding along unverified under the AI label. The MRO sector’s certification machinery is healthy and fast. The AI layer sitting on top of it is still waiting for its own clock to start.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The core argument—that conventional aerospace certification (Part 145, STC) is mature while AI/ML verification standards lag behind—is coherent and well-supported by the cited MRO examples, though the
Source & Claim Verification	Qwen · hosted fallback (qwen-2.5-72b-instruct)	cleared. All factual claims are supported by citations, but the article could benefit from more specific references to the whitepaper and the DO-178C life cycle model.
Regulatory & Framework Fidelity	Mistral · fallback (mistral-medium-3.1)	cleared. The briefing accurately identifies the DO-178C gap for AI/ML in aerospace but fails to explicitly map this to ISO 42001, NIST AI RMF, or other non-aerospace frameworks where relevant.
Technical Accuracy	Llama	cleared. The technical distinction between deterministic DO-178C software verification and the non-deterministic nature of neural network weights/drift is accurately represented.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and contrasts conventional certification with the emerging, less defined ‘AI layer,’ clearly flagging potential vendor hype and the absence of counterarguments.
Novelty & Non-Duplication	Grok	held. The two-clocks contrast is a competent same-week synthesis of commodity AIN MRO items plus a generic DO-178C/ML gap, but the underlying thesis is industry-commonplace and the piece does not demonstrat
Validation	DeepSeek	cleared. The central claim that conventional certification pathways are robust while AI/ML verification lacks equivalent standards is strongly supported by industry evidence and regulatory reality.

Sources cited: 15. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.