Anomaly detection, predictive maintenance and quality monitoring on your real sensor data — running on-prem, deployable down to a 2 EUR microcontroller. The founding cohort gets one full year of Luviner, free.
Full access for 12 months. No credit card. No usage cap. No hidden gating after the first model.
Anomaly detection, predictive maintenance, multivariate quality monitoring, classification. AutoSelect picks the best one for your data, expert override available.
Compile your trained model to a C binary for any target: x86 server, industrial PC (ARM/x86), or microcontroller from 2 EUR upwards. No cloud required at runtime.
Email line directly to the founder for technical questions. Not sales calls, not demos — real help when you get stuck on data shape, training, or deployment.
Your sensor data never leaves your infrastructure. Train and deploy entirely inside your network if needed. We do not see your data.
Five-minute structured feedback at day 7, day 30 and day 60. Helps us improve, helps us tell whether you are stuck.
If the pilot delivers measurable value, we publish a written case study with your numbers, your logo and one quote. Reviewed and approved by you before going live.
At the end of the program, regardless of outcome. Honest conversation about what worked, what did not, what would have made the difference.
Founding partners are publicly listed on the Luviner website with company name and logo. This is the part that builds trust for the next manufacturers we talk to — and it gives you visibility as an early adopter of edge AI in production.
We expect to receive far more applications than spots. To be fair to everyone, here is how we will choose:
Want to see what a CSV looks like? Download a real sample (vibration sensor) · CSV Data Format
Luviner already runs on public benchmarks (SMD, SWaT, SMAP/MSL) with strong results. But synthetic benchmarks do not pay rent for anyone. Real factories have messy sensors, gaps in data, unexpected failure modes, and deployment constraints we cannot simulate from a laptop. This program puts Luviner on five real production lines, learns from what breaks, and turns the surviving installs into the case studies our next 50 customers will read.
Really free for 12 months, full access. The exchange is the three written check-ins, the final interview, and a case study only if the pilot works. After 12 months you decide whether to convert to a paid plan with a founder discount, or walk away with no obligation.
No. You can train and deploy entirely on-premise. We do not require uploading raw sensor data to our servers. The compiled binary runs locally on your hardware. We never see your data unless you choose to share it.
Three options: convert to a paid plan with a permanent founder discount, stay on the free tier with limits, or stop using Luviner with no penalty. There is no auto-renewal, no credit card on file, no surprise invoice.
No. The program is open globally. We will run check-ins in English or Italian. Time-zone overlap with Europe makes communication faster but is not a requirement.
We move you to the waitlist. The first cohort is small on purpose, so we can give every partner real attention. As soon as we open the next round we contact the waitlist first, before publishing anywhere. Filippo Groppi, founder of Luviner, personally reviews every application and responds within ten days.
Yes. The application form takes five minutes. If your problem does not fit, we will tell you honestly and point you in a better direction. If it might fit, we will discuss it before deciding. Do not self-reject.
Fill in the form below. If anything is unclear, write what you can and we will follow up by email. We respond to every applicant within ten days.