The AI retrofit box
Retrofit without touching the process. On-premise AI finds the use cases that pay off for you.
Actively cooled, compact and built for the control cabinet. The box runs continuously right at the machine.
An edge computer with 8 GB RAM and a DAQ card with 8 analogue channels at up to 100 kS/s. Enough for acoustic emission, vibration and process signals.
Mounts on a standard 35 mm DIN rail. The sled carries the electronics and slides out for service.
A Y-adapter taps the signals on the existing cable. Your controller stays untouched.
Your machines have run reliably for years. Yet up to 90 % of them are not connected. The data that decides wear, quality and downtime is lost every single day.
70–85 %
of AI initiatives fail on data access or high analysis costs, or end as a pilot without proven value.
10–20 years
is the typical age of machinery in mid-sized manufacturing. Replacing it rarely pays off.
No data science team
and no budget for in-house development: that is the norm for SMEs with 10 to 500 employees.
Figures based on industry studies on digitalisation and AI in German SMEs.
Instead of months of consulting, our AI pipeline automates the search for use cases. You get a reviewed study and a working prototype.
Day 1
Box on the DIN rail, Y-adapter into the existing cable. No changes to controller or process.
Weeks 1–3
High-frequency sensor signals plus machine data via OPC UA and MQTT. Every measurement is stored locally and mirrored.
Weeks 3–6
Proven statistical methods analyse the data automatically. A local language model derives use cases and rates feasibility and business value.
Weeks 6–8
You receive a reviewed feasibility study and a working prototype, for example for wear prediction.
The entire AI analysis runs on your premises. No cloud, no upload, no loss of know-how.
Measurement data, models and the language model run on hardware in your network.
Nothing is transferred to third parties. That removes the biggest barrier to AI in mid-sized companies.
Process parameters and insights never leave the plant and remain your competitive edge.
We automate what costs weeks of manual work in conventional projects.
| Conventional data science consulting | Opti-Solutions | |
|---|---|---|
| Data access | Integration with the controller, often invasive | Non-invasive retrofit on the existing cable |
| Analysis | Manual work by data scientists | Automated by an agentic AI pipeline |
| Data storage | Often in the cloud | Fully on-premise |
| Outcome | Report with recommendations | Reviewed study plus working prototype |
| Duration | Several months | 6–8 weeks |
Retrofit instead of replacement: upgrading existing equipment costs roughly 30 to 50 % of the investment for a new machine.
Detect tool and grinding wheel wear before it costs quality. Plan changes instead of reacting.
High-frequency acoustic emission signals reveal contact, engagement and irregularities in real time.
The AI learns your machine's normal behaviour and flags deviations long before an operator notices.
Infer part quality from process signals and reduce inspection effort where it matters.
In the field
At an industrial partner, we placed the box between the existing sensor cable of an acoustic emission monitoring system. It reads the machine's start and stop signals and records every measurement automatically. A one-week endurance test confirmed gap-free recording and synchronisation.
Status: working prototype. The industrial version with CE marking and EMC testing is in preparation.
Opti-Solutions builds on ARTHUR, a distributed measurement system developed at Furtwangen University in the research project DQ-Meister and refined in PräziLoop. We are turning that experience into a product for mid-sized manufacturers.
Co-founder, hardware and pilot projects
PhD candidate at Furtwangen University. Industrial retrofit, IIoT and development of the retrofit box.
Co-founder, AI pipeline and sales
Furtwangen University alumnus. Data analysis, local language models and customer projects.
We are starting with a small group of manufacturers in southern Germany. Tell us about your machine and we will get back to you with an initial assessment.
kontakt@ihre-domain.de
or call us [[Telefonnummer]]