Frequently asked questions

With a free, 45-minute discovery call, where we ask about the defect, the asset or the building and what it costs when things go wrong. If there is a fit, the next step is usually a 3-week Feasibility Sprint or a 6-week Proof of Value. How we work describes each stage.

For inspection, 30–50 good parts and whatever defective parts you have; even 5–10 per defect type is useful. For monitoring, an asset list and your recent failure history. We image samples in our optics lab and deliver a written go/no-go with modeled ROI.

A Feasibility Sprint takes 3 weeks and a Proof of Value 6 weeks. Deployment typically takes 6–12 weeks per line or site, covering installation, integration, FAT and SAT, operator training and handover. You receive a written proposal within 5 business days of a scoping call.

Sometimes. Condition monitoring on standard motors, pumps and fans is well understood and can go straight to deployment. For inspection of a new defect, we recommend at least a Feasibility Sprint, because whether a defect can be detected reliably is a question for your samples, not for a sales call.

Because one of our principles is plain answers, and that includes what things cost. Every service, subscription and support tier is listed on our pricing page, so you can build a budget before you talk to us.

It is $9,500, fixed, for 3 weeks. The full amount is credited against a project or subscription started within 90 days. The 6-week Proof of Value is $24,000, with loaner hardware included, and 50% is credited to a subscription.

Inspection-as-a-Service starts at $2,950 per station per month on a 36-month term, with Lensnodes, Gatepoint, Console, Care Essentials and hardware replacement included. A capital purchase is typically $48,000–$140,000 installed for a single station, with Care optional from $650 per station per month. The inspection ROI calculator compares both.

Yes. Once a Proof of Value has established a measured baseline, part of the fee can be tied to measured scrap, escape or downtime reduction against it. Our article on outcome-based pricing explains when that fits.

No. Vision Retrofit adds AI to the cameras you already own, and we select compute across NVIDIA Jetson, Hailo and Qualcomm platforms to suit the job. Our own Lensnode, Pulsenode and Gatepoint hardware is the default where it fits, because we know how it behaves in the field.

Lensnode cameras have IP67 aluminum housings, with an optional IP69K 316L stainless housing for washdown, and operate from −20 to 55 °C. The Gatepoint XT runs from −40 to 70 °C for outdoor, substation and cold-storage sites. The Pulsenode W is IP68, with a Class I Division 2 variant. See hardware.

Subscriptions include hardware replacement. Advance hardware replacement takes 3 business days on Care Essentials and the next business day on Care Verified and Care Dedicated, and Care Dedicated adds an on-site spares kit. If a Gatepoint faults, it fails to a safe state you choose in advance.

Yes. For equipment makers we design custom optics, sensor nodes, embedded Linux, firmware and PCBs. See embedded engineering and smart optical sensors. We also support readiness work for the EU Cyber Resilience Act.

You do. Hardware you buy, every image and measurement, and every trained model belong to you, and models can be exported in the open ONNX format. If you stop working with us, your data and models go with you.

No. Inference, storage and Perflens Console run on the Gatepoint or an on-premise server. Private-cloud aggregation for multi-site fleets is optional, and remote support uses an outbound connection that you control and can switch off. See air-gapped AI.

Devices are hardened in line with IEC 62443, with secure boot, TPM 2.0, signed A/B updates and SBOMs, and they sit on network segments you define. The Security & Trust Center has the details, and our Edge & OT Security service extends the same practices to your wider OT network.

Inspection cameras are framed on product, not people. For occupancy in buildings, counting happens on the device and only numbers leave it, with no images stored or transmitted and no facial recognition.

That depends on the defect, the optics and the line, which is why we measure rather than promise. A Feasibility Sprint estimates detection and false-reject rates on your samples, a Proof of Value measures them on your line against criteria agreed in writing, and every model's model card records its measured performance and known limits.

We combine anomaly detection, which learns mostly from good parts, with supervised models for the defects you do have. Where real examples are scarce, synthetic defect data can broaden training, validated against real samples before we rely on it.

Perflens Care monitors confidence, reject patterns and image statistics against an agreed baseline, and flags drift before detection drops. Retraining runs twice a year on Care Essentials, quarterly on Care Verified and continuously, as needed, on Care Dedicated.

No. Deterministic inspection models make every pass/fail decision. The private copilot runs on a Gatepoint G8 with no cloud, and helps operators ask questions of camera feeds and inspection history, triage alerts and draft reports.

Care Essentials is included with every subscription, and covers drift monitoring, a monthly health report, retraining twice a year and Severity 1 response within 4 business hours. Care Verified adds analyst verification of alerts and 1-hour Severity 1 response; Care Dedicated adds 24/7 coverage, 30-minute Severity 1 response and a named reliability engineer. See support and SLAs.

Our team works Monday to Friday, 8:00 a.m. to 6:00 p.m. Central Time, at (470) 784-4045. Care Dedicated customers have a 24/7 critical line on the same number, which routes after hours.

Yes, anywhere in the contiguous US. Our field engineers travel nationwide; on-site response is within 2 business days on Care Verified and the next business day on Care Dedicated, and scheduled on Care Essentials.

Our headquarters and engineering lab are at 6425 Lakeover Road, Suite C, Jackson, MS 39213. We work with commercial and industrial clients across the United States. About Perflens tells the story of why we started.

Yes. Integration with your PLCs, MES, historians and CMMS is part of every project, not an extra. We connect over OPC UA, MQTT with Sparkplug B, Modbus, EtherNet/IP, PROFINET or REST, and document every connection for your controls and IT teams. See integrations.