AI for manufacturing companies
I'm a design engineer. For over 10 years I've been developing products, from the control system of a flying car to climate control dials for Škoda. I train teams on site, using your own RFQs and drawings.
Where the time goes
Training
We work with your RFQs and drawings, not examples from the internet. Your team tries the method on last week's RFQ and then uses it on every one that follows.
I train job shops and machine builders with 20 to 150 people who do laser cutting, bending, machining, welding or one-off production.
For estimators, process engineers, design engineers and quality. We spend the whole day on your RFQs, drawings and work instructions, and people leave with templates they can use the next day.
Book a call →For owners, managing directors and operations directors. Using one of your RFQs and one of your parts, I show what AI handles well and where it gets things wrong. You leave knowing whether and where to start, and what rules to set for company data.
Book a call →We follow one RFQ from the first email to the quote that goes out and time every step. You'll see where the time goes and what can be faster.
Book a call →A longer course for teams who want to use AI every day. Czech companies can have part of the cost covered by the Labour Office programme “Vzdělávání pro firmy”. The company files the application itself.
Book a call →What AI can and can't do
Where AI helps in metalworking, where it makes things up, and what data to keep out of it.
RFQ to spreadsheet
Turn an email and a PDF drawing into a table of parts, materials and quantities. AI flags what's missing and drafts a question for the customer.
Drawings, CAD and costing
Get costing parameters such as flat size, bends and cut length from a drawing and a STEP or DXF model. I demo it in Chainly.
Work instructions and know-how
A drawing and a few sentences from the process engineer become a draft routing and traveler. An interview with an experienced foreman becomes a guide for new hires.
Complaints and 8D
Turn complaint emails and photos into an 8D report outline. Quality then confirms the root cause.
Rules and next step
We write a one-page AI policy and you pick three tasks to start trying the following week.
Chainly
An RFQ often has dozens of parts, each with a drawing, a model or a line in the BOM. Chainly goes through them for you. For sheet metal parts it measures the flat size, bends, cut length and pierces, and prices the part with your rates. Your estimator checks the numbers instead of retyping them.
See Chainly →Background
Control logic and safety interlocks for the car-to-aircraft transformation.
Read more →
Multifunction control dials for climate and other vehicle functions, shipped in new-generation vehicles.
Read more →
A robotic, remote-controlled drive for year-round autonomous operation.
Read more →Product development
I help you develop your own hardware product, from a five-day audit to a long-term engagement.
In 5 days I review your product and give you a written summary of what's ready for manufacturing, what isn't, and what to fix before you order the first batch.
Book an audit →One day with the founders, working on your product. You leave with a development plan, a spec and a shortlist of suppliers.
Schedule a workshop →A few days a month I work as your head of product development. I join your meetings, work through engineering questions with the team and keep suppliers on track.
We develop your product from first concept to series production. I lead the project and bring in engineers and manufacturers I've worked with before.
Get a quote →Just need a one-off consultation? An hour costs €100.
Book a call → About
I worked on the control system of the AeroMobil flying car, climate control dials for Škoda and a solar-powered pool cover drive for Albixon. I know how much work goes into a single RFQ and how easily a note on a drawing gets lost. That's why I now teach manufacturing companies to use AI on RFQs and drawings, and why I'm building Chainly, which turns RFQ documents into a quote.
Book a 30-minute call and bring one RFQ with drawings. I'll show you what AI could prepare from it and what it couldn't. Then you decide whether training makes sense.