erp2.ai
Manufacturing Execution System (MES) · Poland
A lean MES layer for the shop floor: capture time and work events, monitor with video, then analyze with AI for recommendations — and give workers a knowledge base built from procedures and operational documents (RAG).
LinkedIn headline
Head of R&D at Taqtile.com · Scaling Lean Manufacturing Execution with AI & RAG · Founder of 5scontrol.com · EAM · MES · Video-Based Factory Intelligence
For search and product language that means: MES first, then time tracking and video monitoring as the base signals, then AI analysis & recommendations, and a knowledge base for workers from real shop-floor documents.
By the numbers
Figures from active and pilot manufacturing workflows. Company names and operational data are anonymized.
The problem
Plants often have ERP, cameras, and timesheets — but not one execution picture: who worked which job, how long, what the video shows, and which procedure applies now. Supervisors still chase status in chat while docs sit unused in folders.
A Manufacturing Execution System closes that gap: time tracking and video monitoring feed the board; AI turns events into recommendations; a document knowledge base answers workers on the floor.
About
I'm Siarhei Fedarenka — Head of R&D at Taqtile, founder of 5S Control, with English in Games Sp. z o.o. in Warsaw. I scale lean manufacturing execution with AI & RAG across EAM, MES, and video-based factory intelligence.
On erp2.ai the product story is explicit: a Manufacturing Execution System where time tracking and video monitoring are the base, then AI analysis and recommendations, and a knowledge base from documents so operators get the right procedure in context — plus field products like HVAC Tech Service.
Work starts with the real process, not the stack: which events to capture, which cameras matter, which docs workers need, and which recommendation changes the next shift.
MES stack
Work orders, job sheets, stations, blockers, and live status beside ERP — the execution layer that answers “what is running now?”
Job and station time, attendance against work, and event timestamps that feed load, payroll-adjacent reporting, and AI analysis — not a disconnected timesheet silo.
Video-based factory intelligence via 5S Control: use existing cameras for process and safety signals tied to the same MES events.
Analyze time + video + execution events; surface recommendations; retrieve procedures from a document knowledge base (RAG) so workers get answers at the station.
Use cases
Manufacturing examples from operational environments, plus a shipped product use case for field teams.
Roughly 180 work orders per month across a dozen stations, with statuses from queued and in-work to blocked (materials wait) and done. Paper and chat handoffs were replaced with a live board: steps, kit status, owners, and shift due times.
Outcome: supervisors see blockers and station load without walking the line for every update.
400 tools in 27 categories (~230 check-outs / month) plus 214 FG SKUs. Check-outs, receipts, shipments, and write-offs post to production channels; weekly summaries show movement without full recounts.
Outcome: tool location and sellable FG stay trustworthy through daily events.
A mobile HVAC calculator for pros: hvactechservice.com — scan walls → 3D grid → pick equipment → live estimate → PDF for bids. Same idea as lean execution: shorten the path from field event to a usable decision artifact.
Outcome: estimates in minutes on-site, without spreadsheet gymnastics.
Factory intelligence
Base layer: execution time tracking and 5S Control video monitoring on the same process map. Next layer: AI analysis that recommends actions and feeds a worker knowledge base from SOPs, work instructions, and plant documents — so answers show up where the work happens.
Interfaces
Statuses, stations, quantities, owners, blockers, tool movement, and upcoming decisions — not office AP tables.
Mock UI with anonymized manufacturing scenarios. Replace with live screenshots when approved.
Engagement formats
One or two sessions: stations, events, blockers, who needs the next signal, and which first control loop is worth shipping.
Fast event loop for work orders, tool moves, thresholds, and notifications — before committing to a custom UI.
Roles, statuses, files, history, mobile-friendly boards, and integrations with ERP / chat / video.
I do not replace ERP for the sake of replacement. If accounting and planning already work, I add the execution layer that closes the shop-floor visibility gap.
Approach
Stations, handoffs, kits, tools, and the manual reconciliations that keep production moving.
Statuses, blockers, owners, thresholds, exceptions, and the questions supervisors actually ask.
Shortest path from event to board update — with real operators from day one.
Audit trail, overdue work, load slices, procedure retrieval, and video signals after the base loop works.
Grow only what helps the line, warehouse, maintenance, and plant leadership.
Who it's for
When to reach out
Contact
Message me on LinkedIn. Say which layer you need first: MES board, time tracking, video monitoring, AI recommendations, or a worker knowledge base from documents.
LinkedIn: Siarhei Fedarenka · 5scontrol.com · hvactechservice.com