Edge first
Decisions are made where the data is born. The cloud is optional — useful for looking across a fleet, never required for a reject gate to fire.
About Perflens
Built by engineers from the plant side
Perflens designs and deploys edge AI hardware, computer vision diagnostics and industrial IoT monitoring. We learned the work on night shifts and commissioning benches, and we build for the people who still work there.
Our story
Why Perflens exists
Perflens was started by engineers who had spent their careers on the plant side: commissioning vision systems, chasing bearing failures on night shifts, and writing firmware for sensors that had to survive washdown.
They kept seeing the same pattern. Quality, maintenance and facilities each bought separate systems from separate vendors, and none of them talked to each other. A vision system rejected parts without knowing the line had just changed speed. A vibration platform flagged a motor without knowing it drove the station that was rejecting those parts. And most of the intelligence lived in a cloud the plant didn't control, so when the internet link dropped, the insight dropped with it.
Perflens exists to put one coherent layer of sensing and judgment at the edge, on hardware the client owns. The name says what we do: perf for performance, and lens — measuring performance through optics and sensing.
Our mission
Make every machine observable, and every observation useful
The first half of that sentence is engineering: the right sensor in the right place, with the right light or the right mounting, surviving its environment for years.
The second half is harder. An observation is only useful if it reaches the person who can act on it, explains itself, and is right often enough to be trusted. Plenty of monitoring projects get the first half right and stall on the second. We hold ourselves to both.
Principles
Five principles we work by
You own it
Hardware, trained models and all data belong to you. Models export as ONNX and data exports in open formats, so there is no lock-in.
Measured, not promised
Every engagement starts with a baseline and ends with the same measurement. If we can't measure an improvement, we don't claim it.
Boring reliability
Industrial enclosures, watchdogs, A/B updates and graceful degradation. Exciting technology, boring uptime.
Plain answers
Every alert explains itself, every model has a model card and every price is published.
What's different
Three things we do differently
One integrator for the line, the machines and the building. Inspection, condition monitoring and facility operations usually arrive as three projects with three vendors. We cover all three with one Edge Stack, one Console and one support desk, so a reject trend, a gearbox alert and a compressed-air baseline can be read side by side.
Edge-first architecture. Our systems decide on hardware beside the asset and keep working without an internet connection. That is a design choice, not a deployment option; latency, privacy and resilience all follow from it.
Published prices. Our pricing is on this website, from a $9,500 Feasibility Sprint to subscriptions and capital purchases. You can decide whether a conversation is worth having before you have it.
Where we work
An engineering lab, and engineers on the road
Our headquarters and engineering lab are at 6425 Lakeover Road, Suite C, Jackson, MS 39213. The lab is where projects start and where much of the proving happens:
- The optics lab, where samples from every Feasibility Sprint are imaged under several lighting geometries and camera setups
- The hardware and firmware benches, where Lensnode, Pulsenode and Gatepoint are designed, built and tested
- The acceptance area, where inspection cells run their factory acceptance tests before they ship
- The Care desk, where analysts watch model health and verify alerts for customer sites
Our field engineers travel to commercial and industrial sites across the United States for surveys, installation and acceptance testing.
More about us
The people and the technology
Leadership and specialists
The eight people who lead engineering, field work, security and care at Perflens.
Read more
Careers
Five open roles for engineers, analysts and technicians who like hardware in the real world.
Read more
Technology we build on
The vendor-neutral compute, runtimes, libraries and protocols inside the Edge Stack.
Read moreFAQ
Questions we hear often
Is Perflens a reseller?
No. We design our own hardware — Lensnode, Pulsenode and Gatepoint — around commercially available compute from NVIDIA, Hailo and Qualcomm. We choose per application, and we have no partner agreements that reward one choice over another.
Where do you work?
Across the United States. Engineering happens in our lab, and field engineers travel to customer sites for surveys, installation, acceptance testing and on-site support.
Do you work with equipment makers?
Yes. Engineering teams at machine builders use our embedded engineering and edge AI hardware integration services to put vision and AI inside their own equipment.
Talk to the engineers who would do the work
A discovery call is with an engineer, not a sales script. Bring the defect, the asset or the building that is costing you most.