Rigolleau S.A.
www.rigolleau.com.arRigolleau brings real-time visibility to a century of glass
A 14-month modernization across furnace, packaging, and pharmaceutical lines — replacing siloed PLC logs with a single OEE feed shared by operators, planners, and quality.
Vision
Give every shift across three distinct product lines the same operational picture — without ripping out a working factory floor.
Impact
A single operating layer that connects the furnace, packaging, and pharmaceutical clean rooms, with defect rate down 31% and changeover time cut in half.
Key services
- Cloud & Platform Engineering
- Data & AI
- Custom Software
- Experience Design
Customer results
- −31%
- Defect rate, year over year
- 2×
- Faster product changeovers
- 14 months
- Kickoff to platform GA
The challenge
Three lines, three operating systems, one factory floor
Rigolleau runs three deeply distinct businesses out of the same campus. Household glassware sells through retail chains and competes on price per pallet. Food and beverage containers ship to bottlers whose schedules shift weekly. Pharmaceutical ampoules and vials live under regulatory scrutiny most consumer plants never see. For most of the company's history, each line ran on its own systems — PLC logs in one place, quality sheets in another, planning in a third. The information existed; it just never met in one room.
What pushed the conversation from "we should fix this" to "we have to fix this" was demand. A new line of pharmaceutical ampoules booked through Q3, a major retailer reshuffling its tableware program, and a contract for amber food containers all landed inside the same six weeks. Without a shared operational view, every changeover became an argument. Every quality escalation lost a day in handoffs.
Our approach
One feed, three audiences
We started by listening on the floor. Two engineers and a designer spent three weeks at the plant — shadowing shifts, sitting in planning meetings, and watching the moment a defect was caught and routed for rework. The pattern was clear: each role needed the same data shaped differently. Operators wanted real-time OEE on the line they were running. Planners wanted predicted finish times. Quality wanted defect lineage from raw material to pallet. Leadership wanted week-over-week trend lines without having to ask.
So instead of pitching a monolithic MES, we proposed a single operational data layer with three distinct surfaces on top of it — each tuned to a role, each pulling from the same source of truth. We sequenced the 14-month rollout against the line schedule rather than against a Gantt chart of our preference: the busiest line went last, after the platform had been proven on a lower-stakes one. That phasing turned out to be the most important design decision we made.
“We didn't need new shifts or new machines. We needed all the shifts to see the same plant.”
What we built
The plant's own operating system
We built the platform in three layers, each addressing a different time horizon and a different audience. The result is one stack that operators, planners, and leadership use every day — and that quality and regulatory can audit on demand.
A real-time floor layer
At the furnace and packaging lines, we stood up a streaming layer that ingests PLC events, vision-system feeds, and operator inputs from rugged tablets at each station. The operator surface — built mobile-first, even though it lives on fixed displays — shows OEE, current SKU, and any open quality holds for the exact shift. Operators don't have to leave the line to log an issue; the same form that captures the issue tags it with the upstream event that caused it.
A planning and capacity layer
Above the floor layer, we built a planning surface that projects line capacity eight weeks out using historical changeover data and the current sales pipeline. Planners no longer rebuild the schedule from scratch every Monday; they accept, edit, or override a recommended plan. The recommendation engine is intentionally explainable — every suggested changeover comes with the three constraints that drove it, so planners can argue with it (and frequently improve on it).
A quality and regulatory layer
For the pharmaceutical line, we built a separate clean-room surface that meets GMP traceability requirements. Every ampoule batch is linked to the furnace temperature curve, raw material lot, and operator signoff at the moment of inspection. Regulatory audits that used to take a week of warehouse archaeology now complete in two business days, with a single export. The same lineage data powers root-cause analysis when a defect escapes into the field — closing a loop that used to take months.
Customer impact
Smaller, calmer, faster
Fourteen months after kickoff, the platform was running every product line at the campus. Defect rate dropped 31% year over year — most of it from catching upstream furnace anomalies before they propagated. Product changeovers, the single biggest source of lost production hours, halved. Planning meetings — historically a two-hour weekly negotiation — became a 25-minute alignment around a shared screen. Quality holds that used to ricochet between four supervisors now close on the same shift they're opened.
The numbers matter, but the harder change is cultural. Operators trust the data because they can correct it. Planners trust the recommendations because they can edit them. Quality trusts the lineage because they helped design it. None of that shows up on a dashboard, and all of it shows up in the next year of production.
What’s next
What's next
With the operational layer in place, the next phase moves into predictive territory: furnace yield modeling, dynamic raw-material allocation, and a customer-facing portal that gives bottlers and pharma clients real-time order visibility. Each of those builds on the same data layer rather than starting over — which was the point all along.