drukomat

We built a Brain for pricing the unpriceable

We transformed Drukomat’s manual quoting into a smart, explainable pricing engine that codifies expert knowledge and scales effortlessly across products.
+250
handled variables
>95%
estimate accuracy
528
logic steps

Driving efficiency

Confidence in Accuracy
Algorithmic pricing now mirrors expert estimates with close to 100% accuracy for simple quotes, while advanced offers show only a 1–5% deviation – reflecting the final layer of optimization that senior estimators would normally add.
Dynamic Quote Generation
A fully functional E2E system capable of replacing manual pricing for non-standard and out-of-standard offers, with dynamic quote generation across 13 complex product types.
Instant Historical Search & Reuse
An offer search engine that lets users instantly retrieve historical quotes based on key attributes – and soon, by similarity.

From expert knowledge to digital process

Drukomat, one of the largest print eCommerce platforms in Poland, faced a fundamental obstacle: manual quoting didn’t scale. Each custom quote relied on experienced estimators and expert knowledge built over years of practice, captured only through large, complex Excel sheets.
The current process depended on multiple disconnected sources of information – paper prices, machine specifications, process calculations – scattered across spreadsheets and systems. Much of this knowledge lived only in the minds of senior estimators. If one left, the company risked losing critical know-how, while onboarding new staff was slow and resource-intensive. As the number of offers and requests grew, so did inconsistency, inefficiency, and time-to-quote.

Not to replace people, but to support them.

The goal was to build a system that assists estimators: automating routine quotes, codifying expert knowledge for training, and freeing specialists to focus on complex, high-value offers where advanced optimization matters most.
Previous industry attempts had failed due to the sheer complexity of non-standard quoting: the growing number of decisions, ever-changing machine specifications, shifting process rules, and expanding web of dependencies.

Drukomat therefore needed not just automation, but a smart and explainable decision-making engine: one that makes processes transparent and adjustable, reduces the time and number of people involved, and reuses past quotes for greater consistency. A system that sales teams could trust. A digital colleague that mimics the way experts think – only faster.

From manual know-how to machine reasoning

We built an AI system that acts as an estimator – a “digital colleague” that automatically completes the same Excel sheets a human would, powered by codified expertise and transparent logic. After the successful migration, we worked with the client to rethink what the platform could become – beyond static content delivery.

Human expertise meets machine precision

Breadth
13 product types
fully automated
Transparency
Visual decision trees for training and auditing.
Horizontal Coverage
We achieved complete architecture coverage: the entire flow from input to price output works E2E for 13 of the most frequently sold products.
Decision making system engine
The quote engine is a visual, traceable process showing decisions taken on:
  • Paper & format
  • Machine selection
  • Print run
  • Folding/binding
  • Packaging
Vertical Depth
Accurate quoting logic across product variants: from glued vs. stitched catalogs to folded vs. flat leaflets. Each path is explicitly mapped and testable.
expertise

The client’s opinion

“The speed and quality of delivery exceeded my expectations. We’re now automating quotes that used to take hours, with a level of transparency that makes it easy to trust and improve the process. It gives us a strong foundation for scaling.”
Marlena Kaźmierczak
E-Commerce Director @ Drukomat

What's next?

Simplified UI

Dynamic form inputs tailored to each product type will make the quoting process faster and more intuitive.

Expanding to new product groups

The foundation is already in place, and new types can be added iteratively as business needs evolve.

Confidence layer

Historical data will be used to flag anomalous quotes before they are sent, strengthening trust in automated results.

Refined offer similarity search

Switching to fuzzy retrieval ensures consistent quoting, simplifies offer creation, and enables quick access to historical comparisons.

Customer-facing interface

As system accuracy is validated, product groups can be standardized and offered directly to clients, eliminating the need for estimator in routine cases.

Process optimization & dynamic pricing

With control over quoting logic, Drukomat can optimize machine use, production planning, and margins, while dynamic pricing adjusts offers in real time based on workload, demand, or stock.

Ready to turn your most complex, manual processes into a scalable automation engine?

Book a free consultation today