# Percepto Axion Opens AI Inspections to Third-Party Drones and Sensors

The new inspection service separates asset intelligence from drone ownership, but compatibility and validation still matter.

By Paras Madan | 2026-10-28


Percepto launched Axion on October 27, a software service that analyzes infrastructure inspections collected by supported third-party drones and sensors. Operators can start with existing data rather than buying Percepto aircraft, making the release relevant to utilities and industrial teams with inspection programs already running.

The useful change is the buying decision. A company can evaluate the analysis layer separately from its choice of aircraft, camera or data-collection contractor.

## What Percepto Axion does

The [launch announcement](https://percepto.com/percepto-launches-hardware-agnostic-axion-platform-bringing-physical-ai-to-mission-critical-operations/) describes Axion as available now through a software-as-a-service offering. Percepto says it uses more than 160 specialized AI models trained on over 100,000 hours of autonomous inspection data.

Its own Ranger drones, Harbor stations and remote operations remain available when customers need additional collection. They are optional ways to supply observations, rather than the only route into the new service.

**Hardware-agnostic does not mean every device or file is supported.** The announcement consistently qualifies third-party inputs as supported sources. It does not publish a complete compatibility list or public price.

## How the inspection AI works

The [Axion product FAQ](https://percepto.com/axion/) describes a combination of vision-language scene understanding, customized Segment Anything component recognition, specialized defect detection and a language-model assistant. Inputs include RGB images, radiometric thermal data, optical gas imaging video and orthomosaic maps.

Findings are organized against GIS-referenced assets, with earlier inspections providing context. Inspection experts review and validate the AI findings. Percepto also lists GET APIs and customized integration options for connecting results to enterprise systems.

That human review is important. A suspicious patch in a thermal image is not automatically a maintenance order. The operating team needs to know which asset was observed, when it was observed and what evidence supports the finding before it changes a work schedule.

## What existing inspection teams should check

Start with a representative historical dataset, not the cleanest demonstration footage. Ask whether Axion can associate observations with the correct asset when cameras, contractors and naming conventions change between inspections.

Then test the handoff. Can a reviewer trace a finding back to its source? Does the receiving maintenance system retain the asset identifier and supporting evidence? How are disputed findings corrected before they appear in a later comparison?

These are evaluation questions, not claims that Axion fails them. They matter because accepting more data sources can increase coverage while also increasing inconsistency. A model's ability to flag a defect is only one part of a usable inspection workflow.

## The opportunity is in the existing archive

An operator with years of images may have useful evidence that is difficult to compare because it lives in separate reports. Axion's approach makes that archive a plausible starting point, instead of requiring a new fleet before analysis begins.

The next test is whether mixed-source observations become reliable asset history. That is the difference between generating more inspection alerts and helping a team decide what to repair next.

I'm Paras. I work at the intersection of AI, engineering and distribution.

I help technical teams turn complex AI products and launches into clear stories that developers and founders can understand and share.

[Book a 30-minute call](https://cal.com/paras-madan/30min)