Atlas
Core-Process Optimization and Intelligent Production Control
Hold the operation steady.
KPT's flagship Solution. We mathematically model your core production process — variables, constraints, decision logic — then run an optimizer + ML engine against your full data surface. Recommendations flow back to your planners through a UI they can audit, override, and approve before anything writes back to SAP/MES. No SAP customization required.
What you get
A decision-support layer on top of your existing operations stack.
Solution 1 reads from SAP, MES, OPC-UA, MQTT — whatever you already have running — on a cadence you configure. The optimizer + ML engine work the data in a sandboxed compute layer next to your system of record. The output is a set of recommendations, not commands. Your planners review, override, approve. Only then does anything write back.
The flagship Optimization in this Solution is Scheduling for Production Lines — multi-line batch sequencing under changeover, capacity, and service-level constraints. Perfetti Van Melle is running it on 3 production lines in their Lainate plant.
The same engine extends to the other Optimizations on the roadmap below: changeover-sequence minimization, OEE loss decomposition, shift-crew optimization, energy-aware scheduling, multi-line load balancing.
Atlas roadmap
7 Optimizations. 1 live, 3 in development, 3 on the roadmap.
Each Solution ships as a sequence of bounded Optimizations. The live ones can be adopted today; the in-development ones are funded and scheduled; the roadmap ones are scoped, not yet committed. Customers can join at any phase — pilot what's live, co-fund what's in development, or shape what's still on the board.
Available now
1-
Scheduling for Production Lines
Multi-line batch sequencing under changeover, capacity, and service-level constraints. Today's flagship Optimization, running on three production lines.
- Effort
- L
- Impact
- XL
- Status
- Now · PVM — Lainate, Italy
In development
3-
Changeover-Sequence Minimization
Sequence the day's SKU portfolio to minimize total changeover time across a multi-stage line. Reduces non-value-added downtime by 8–15%.
- Effort
- M
- Impact
- L
- Target
- 2026 Q3
-
Energy-Aware Scheduling
Bias production windows toward off-peak energy pricing while respecting service-level commitments. Useful where energy is 10%+ of cost of goods.
- Effort
- M
- Impact
- L
- Target
- 2026 Q4
-
Grade-Change Loss Reduction
Minimize broke + off-spec output during grade transitions on continuous-process lines (pulp & paper, steel, polymers). Optimizes transition timing and sequence to lower transition waste per ton.
- Effort
- M
- Impact
- L
- Target
- 2027 Q1
On the roadmap
3-
OEE Loss Decomposition
Decompose Overall Equipment Effectiveness loss into availability / performance / quality components and trace each loss source back to controllable variables.
- Effort
- L
- Impact
- L
- Planned
- 2027 Q1
-
Multi-Line Load Balancing
Real-time rebalancing of work between parallel production lines when one line drifts (breakdowns, quality holds, material shortages).
- Effort
- L
- Impact
- L
- Planned
- 2027 Q2
-
Shift-Crew Optimization
Right-size shift crew composition (operators, maintainers, quality) per shift pattern based on planned production mix and historical labor utilization.
- Effort
- M
- Impact
- M
- Planned
- 2027 Q3
Effort and Impact are estimated on a S / M / L / XL scale (1 dot to 4 dots). Effort = engineering work required to ship; Impact = expected operational improvement at typical industrial scale. Estimates are KPT internal benchmarks and are validated against your data during the 30-day shadow-run PoC before any commitment to scale.
Where this Solution lands
Industries we've prototyped this for
Confectionery
High-SKU lines + tight changeover windows + thin margins. The combinatorial space is too big for spreadsheets.
PVM (Lainate) — live
See Confectionery →
Beverage
Filling-line sequencing under demand variance + returnable-asset logistics.
See Beverage →
Pulp & Paper
Paper-grade transitions + energy-aware production scheduling.
See Pulp & Paper →
Mining
Asset-intensive scheduling around SAG mill availability + energy peak windows.
See Mining →
Want this on your plant?
Start with one production line.
A 2-week assessment locks the data sources, the optimization model, and the success criteria. PoC ships in 4–6 weeks with shadow-run trust gates baked in. Measurable outcomes in dollars before any commitment to scale.