Grupa KRES Custom AI Pipeline Automates 100 Monthly Stone Orders — Quote Time Cut from 5 Days to 1 Day

Title

Grupa KRES Custom AI Pipeline Automates 100 Monthly Stone Orders — Quote Time Cut from 5 Days to 1 Day

Content

ALGORCOMP built a custom AI-powered order pipeline for Polish stoneworker GRUPA KRES sp. z o.o. (20-person team, ~100 orders/month) that consolidated the 9-stage, 6-role order flow from six disconnected tools into one system — time from first customer inquiry to ready quote dropped from 5–7 days to 1 day, and margin per order rose by 6 percentage points within six months of rollout. "The best thing about this system isn't that we handle more orders — we handle the same number. The best thing is that the same 20-person team now has time to talk to customers instead of looking for a card in a binder. We used to be chaos managers. Now we manage a business," said the Owner of GRUPA KRES sp. z o.o. Before the AI pipeline, every step — inquiry, measurement scheduling, on-site measurement, quote building, approval, workshop order, quality control, transport, installation, final documentation — lived in a different tool (landline, paper notebook, calculation Excel, Google Calendar, paper workshop floor, accounting system). Inquiries queued 5–7 days on a single measurer's calendar, then on a single calculator who knew material prices — and pricing was inconsistent, because the same granite countertop was quoted at different prices by different staff using their own price lists and margins. Total rollout: 14 weeks. Weeks 1–3 — workshops with every role and discovery on 200 historical orders; produced a process map and a pricing knowledge base of 320 stone types, 18 edge-finishing variants, 12 surface-finish types, plus installation-specific formulas. 9 weeks of build. 2 weeks of staged rollout: week 1 every new inquiry ran in parallel through old + new with every AI quote verified; week 2 the new system became primary; week 3 full trust with a 15-minute daily team stand-up to catch edge cases. Architecture: Node.js + PostgreSQL backend with a full order model. AI layer is OpenAI GPT-4o handling three jobs — (1) parses incoming inquiries (phone transcription, web form, email, WhatsApp) and extracts structured project data; (2) generates a preliminary quote from historical orders plus the current material price list; (3) produces a ready quote PDF with scope and delivery timeline. React frontend with four views (daily reception pipeline, measurement calendar, workshop queue, installation schedule) plus a mobile app for installers handling on-site photo documentation. Operational scale: 100 orders/month in one unified pipeline (up from 6 disconnected tools). 0 orders lost post-rollout. 20-person team. Pricing knowledge base: 320 stone types, 18 edge-finish variants, 12 surface-finish types. Time-to-quote: 1 day (from 5–7 days). Margin per order: +6 pp within 6 months. Customer NPS: +28 points in Q1 post-rollout. The case study does not describe a stated next-phase roadmap; the source positions the deployment as complete, with the customer portal (status link in the confirmation email) and the 15-minute daily stand-up as the steady-state operating model.