Launching a New Chapter in Vision AI
Last week at the Embedded Vision Summit, Plainsight CEO Kit Merker unveiled OpenFilter, an open‑source project designed to run vision workloads at scale. OpenFilter distills years of Plainsight engineering into a lightweight, high‑performance filtering engine that developers can embed anywhere, from the public cloud to the factory floor, without the bloat or cost overhead that often slows computer‑vision projects.
Joining Kit for the launch was Pankaj Mohindru, Director at BrickRed Systems, a global consultancy with more than 850 data and AI specialists. BrickRed has spent the past two years testing Plainsight’s technology across manufacturing, retail, transportation, agriculture, and food‑quality use cases. Their verdict: OpenFilter cuts server costs and developer ramp up time to a fraction of legacy YOLO‑based pipelines.
Why Developers Are Excited
- Time‑to‑Value — BrickRed engineers moved from concept to working prototype in a single weekend. One developer built a scratch‑and‑dent detection model for car dealerships overnight, thanks to intuitive APIs and off‑the‑shelf annotation data.
- Cost Efficiency — OpenFilter’s lightweight runtime trims GPU usage dramatically. For BrickRed, that translated into fractional server bills compared with heavyweight YOLO deployments.
- Enterprise Readiness — Built‑in observability, secure multi‑tenancy, and single sign‑on mean teams can scale from pilot to global rollout without rewriting code or bolting on extra tools.
- Open Ecosystem — By releasing the core as open source, Plainsight invites the community to extend, audit, and optimize filters—while keeping an enterprise‑grade platform behind the scenes for teams that need support and governance.
Rapid Prototyping, Real Results
BrickRed’s approach is simple: Show, don’t tell. Instead of walking into meetings with slide‑wear, the consultancy now arrives with fully functional demos that speak the language of the customer’s business.
- Car Dealers: A scratch‑and‑dent appraisal tool, built in four days, delivers service quotes in minutes and helps smaller dealerships compete with national brands.
- Food Production: Real‑time dough‑size monitoring prevents 10,000 off‑spec donuts from leaving the line, saving raw materials and rework.
- Industrial Safety: Helmet and vest detection prototypes help manufacturers validate PPE compliance before accidents happen.
The Market Is Ready—and So Is the Hardware
A decade ago, vision AI stalled on GPU limits and camera quality. Today, real‑time inference is table stakes, and enterprises want answers now, not ten minutes later. With stronger edge devices and smarter filters, use cases that once felt futuristic, counting cattle in open pastures or gauging donut diameter on a conveyor, are moving into full‑scale deployment.
OpenFilter arrives at just the right moment, blending community transparency with the guardrails that Fortune 500 security teams demand. As Pankaj put it:
“Enterprise decision‑makers care about security and scalability first. OpenFilter covers both, then lets us innovate on top.”
What’s Next
Plainsight will continue expanding its filter library, while partners like BrickRed bring domain expertise and integration muscle. Together, they’re compressing the old three‑year cycle of ‘plan, pilot, re‑platform’ into a single sprint:
- Define the vision.
- Prototype in hours.
- Deploy with confidence.
Focus on your vision—Plainsight handles the infrastructure.
Ready to get involved? Explore the code at openfilter.io, dig into the docs, and join the conversation on Discord.
