When the shift breaks, the plant doesn't stop.
qb-Plant finds who can legally and safely replace an absent worker. Every option is checked against the collective agreement, certifications, medical fitness and rest rules before the supervisor approves — and every decision is recorded.
At 5:40 AM a shift breaks.
A worker calls in sick, does not show up, or an emergency occurs.
Collective agreement · Working time and rest · Certification · Medical fitness · Production plan
Supervisor · Phone · Roster · Excel · Memory
The supervisor keeps the plant running — and works blind.
Why it is getting harder: fewer qualified workers → fewer eligible replacements for every absence. 77% of EU businesses cite skilled-staff shortages as a barrier to investment · 55% of metalworking hires in Italy are hard to fill.
Sources: Council of the EU, doc. 14896/25 (EIB Investment Survey 2025) · CNEL–Unioncamere, July 2026 · Rules: D.Lgs. 66/2003; D.Lgs. 81/2008
Workforce systems plan the shift. qb-Plant acts when the plan breaks.
At 5:40 a worker calls in sick. The supervisor opens qb-Plant, sees three replacements already checked against contract, certification, fitness and rest, approves one — and the decision is recorded.
The absence reaches qb-Plant.
qb-Plant finds who can legally and safely replace the worker.
The supervisor picks one of three ranked options.
Who decided, on which rule, is saved.
The supervisor decides. qb-Plant prepares the decision and keeps the proof.
Designed as a decision layer above existing HRIS, payroll, time-and-attendance and production systems.
Rules are enforced, not guessed.
qb-Plant runs on small, super-specialised agents. They are deterministic and non-generative: collective agreements, working-time limits, certifications and fitness requirements are encoded as explicit rules, each linked to its article. The same situation always produces the same checks and the same verdict, and no generative model decides who can work.
Hard constraints, always applied — a plan that breaks a rule is blocked before approval.
Same input, same verdict — rule checks and rankings are deterministic and repeatable.
Light by design — specialised agents need few computing resources and run in the EU cloud or on the client's own servers.
Every decision traceable — who approved, when, on which rule and article.
How qb-Plant works — inside the MVP.
Screens from the qb-Plant MVP, built in four weeks on real data from a leading Italian steel group. Worker names are fictitious.
- Supervisor approves — qb-Plant never executes alone.
- Workers notified, confirmations tracked.
- Decision recorded with the rule and article applied.
Plant graph: people · skills · certifications · machines · shifts · rules → hard constraints → agreement rules as data, cited by article → 3 ranked feasible replacements
Same absence. 7 manual steps today, 3 with qb-Plant.
- 1Absence arrives by phone or message
- 2Supervisor checks the paper roster
- 3Calls candidates one by one
- 4Checks contract, rest and fitness from memory
- 5Updates the Excel plan
- 6Informs HR and the team
- 7No record of why that person was chosen
- 1The absence enters qb-Plant
- 2Three replacements, already checked against contract, certification, fitness and rest
- 3Supervisor approves · workers notified · decision logged with the rule applied
Live in the MVP: qb-Plant blocks a plan that violates the encoded weekly-rest rule and cites the applicable article — CCNL Art. 14.
Time saved, decision speed and economic value will be measured in paid pilots.
WFM manages the workforce. qb-Plant focuses on the disruption decision.
- Phone
- Excel
- Paper
- Messages
- Supervisor judgement
Broad capabilities may include:
- Forecasting
- Scheduling
- Absence management
- Workforce deployment
- Shift changes
- Compliance
- Industrial disruption decision layer
- Hard-constraint resolution
- Ranked feasible replacements
- Agreement-to-rule representation
- Human approval
- Rule + article decision trace
- Overlay above existing workforce systems
We do not replace your workforce systems. We focus on the high-friction decision that occurs when disruption hits.
Why now: fewer people, stricter rules.
of EU businesses cite skilled-staff shortages as a barrier to investment · 55% of metalworking hires in Italy are hard to fill → fewer eligible replacements for every absence
AI Act high-risk requirements apply to systems under Art. 6(2) and Annex III, including certain worker-management use cases → higher governance standards for these systems
Built for privacy, traceability and human oversight from day one. — Supervisor approves · Rule + article recorded · Hard constraints enforced · Non-generative, deterministic agents · EU cloud or on-premise · client-controlled data.
Designed for Annex III readiness — conformity programme in the product roadmap.
Sources: Council of the EU, doc. 14896/25 (EIB Investment Survey 2025) · CNEL–Unioncamere, July 2026 · Reg. (EU) 2024/1689, Art. 6(2), 6(3), Annex III pt. 4, Art. 14; timeline as amended by Reg. (EU) 2026/1744, Art. 113.
Plant-led. HR- and compliance-enabled.
- Plant Manager / COO
- CFO where relevant
- HR Director sponsor
- RSPP / safety
- unions
- IT / security where relevant
- Paid pilot on your data
- Deployment & integration
- Recurring software licence
- Multi-site expansion
Steel · Chemicals · Paper · Ports — North-East Italy and Lombardy
Tested on real plant data.
qb-Plant MVP built in four weeks on the dataset of a leading Italian steel group, with a positive assessment from its HR committee.
Signed letter of intent with a rail-port logistics operator.
Built on qb-Agentic, Pragmatica's core of specialised, non-generative AI agents.
→ Explore our agentic AIqb-Plant— when the shift breaks, the plant doesn’t stop.
Next step: a 30-minute call and a live demo of the qb-Plant MVP.