Imagine: you receive a drawing of a new part. Previously, a technologist would spend several days, or even a week, analyzing it, calculating time norms, preparing routes and specifications. Now, this work is done by a neural network in 1–5 minutes. And it doesn't just perform the task; it learns from your production, adapting to your machines and standards.
Today, the webinar 'AI.Prom 2026. Day 2' showcased two such products — 'AI-Technologist' and 'AI-Estimator'. Here’s what they are, how they work, and their costs.
What is 'AI-Technologist' and How Does It Work
'AI-Technologist' is a multi-agent system based on AI. In simple terms: it’s not just one neural network, but an entire orchestra of several, each responsible for its own task. One recognizes the drawing, another calculates time according to GOST standards, and a third generates the technical map. Together, they automate production preparation.
Capabilities:
- Read drawings (even old ones, scanned with stains and creases).
- Calculate time norms for each operation (cutting, welding, machining, painting).
- Create production routes in accounting systems (1C, SAP, any via API).
- Calculate cost and material intensity.
- Generate specifications and technical maps.
- Account for movements between workshops and even outsourcing.
Speed: from 1 to 5 minutes for a standard drawing. A complex multi-page one may take several hours. But that’s still not days of manual work.
Accuracy: after training on your data — 88–95%. In the demo version without training — 40–60%, but that’s just the starting point.
Security and Integration — Key Advantages
The developers focused on local deployment. The system is installed on your server and can operate entirely without the internet. This is ideal for enterprises with requirements under Federal Law 152, commercial confidentiality, and other regimes.
It integrates with 1C (UNF, ERP), with any software via API. You can connect it to CNC machines to receive real-time data. You upload your machine park once, and then the system suggests which equipment is best for producing the part.
Implementation Process and Costs
The process occurs in three stages:
-
Pre-project survey — 180,000 rubles. Lasts 1–1.5 weeks. Specialists study your production, interview technologists and designers, and determine where to upload data. The output is a detailed technical assignment with an implementation plan and guaranteed accuracy percentage.
-
Version selection:
- Basic — like the demo, but locally, calculation according to GOST, without modifications.
- Standard — loading of standards, operation routes, matching with 1C.
- Professional — calculation of material intensity, selection of material analogs.
- Enterprise — full integration with machines, web interface, predictive analytics, integration with any systems. Price — from 3 million rubles, but it greatly depends on the volume of modifications.
-
Implementation cost — determined individually based on the technical assignment. You can break it into stages and stop when you achieve the desired functionality.
Support — 100,000 rubles per year. Includes updates (neural networks evolve quickly, models become lighter and more accurate) and consultations.
Surprise: 'AI-Estimator' for Engineering Systems
Another product showcased at the webinar was 'AI-Estimator'. It was born from a real client request. The company 'ZimaLeto' (manufacturer of air ducts and ventilation installation) previously spent 1 to 5 days processing a single project. Now it takes minutes.
How it works: you upload a PDF of the project (heating, ventilation, air conditioning), the system highlights sections, specifications, matches them with your price lists, and compiles an Excel estimate with formulas. If an item is not in the supplier's price list but you produce it yourself — your internal price list is loaded.
Prices, VAT, linear meters converted to standard segments — everything is automatic. The output is a ready commercial proposal.
The company founder Alexey Narozhny confirmed live: the speed of response has become a real competitive advantage, and the number of successful deals has increased.
Live Questions from the Webinar and Answers
- Does it recognize old Soviet drawings with stains? Yes, it redraws lines. If a parameter is unreadable (for example, roughness coefficient), the system will ask a question — you clarify, and it recalculates.
- Can it be used without the internet? Yes. Completely. Even if your external network is cut off, the program will continue to work.
- Does it account for tool wear (electrodes, cutting tools)? Yes, it is factored into the production context. The system can even generate a report for the warehouse: 'issue this many electrodes to the workshop manager before the shift'.
- What about welding? It selects the workshop, type of welding (TIG, MIG, arc), time. If you have your own modes — you upload them, and it picks them up.
- What about reinforced concrete structures? Not yet, but it can be taught. It all depends on the context.
Conclusion: Who Needs This
'AI-Technologist' and 'AI-Estimator' are not toys for demonstration, but working tools that are already implemented with clients. They solve specific pain points:
- Lack of qualified technologists in regions;
- Long manual processing of drawings and projects;
- Errors in calculating time and cost;
- Complexity of integration with accounting systems.
If your production is related to metalworking, machine engineering, ventilation, or any other drawing interpretation — it makes sense to request a demo version. Especially since the pre-project survey can currently be obtained for free.