Data architecture – “AI can’t scale without it”

At the Smart Manufacturing Food and Beverage Europe 2025 summit in London, Ricard Torralba, artificial intelligence lead at AG Solution Group, delivered a compelling keynote that reframed how manufacturers should think about data architecture. His central thesis? That the unified time space is not just a technical innovation — it’s a strategic imperative for scaling artificial intelligence across industrial operations.

What is the unified time space?
Torralba described the unified time space as “a harmonised layer where all data — regardless of origin — is indexed in time and made contextually accessible.”

This architecture allows manufacturers to unify disparate data streams from sensors, PLCs, MES, ERP systems, and even external sources into a single, time-aligned namespace.

“It’s not about centralising data,” Torralba explained. “It’s about synchronising it — giving every data point a shared temporal context so AI can reason across systems, not just within them.”

Why it unlocks AI scalability
One of the most pressing challenges in food and beverage manufacturing is deploying AI across multiple lines, plants, and systems. Torralba argued that without a unified time space, AI models are trapped in silos, unable to generalise or scale.

“AI needs clean, contextual, and timely data. If your data lives in silos, your AI will too,” he said. “The unified time space breaks that barrier. It’s the difference between having a thousand isolated snapshots and having a live, panoramic view of your operations.”

He illustrated how this approach accelerates model deployment, improves real-time decision-making, and simplifies governance. “When your data is harmonised in time, you don’t just get better insights — you get faster, more reliable action.”

Real-world impact
Torralba shared case studies where the unified time space enabled predictive maintenance, energy optimisation, and adaptive scheduling. In one example, a beverage manufacturer reduced downtime by 30% using AI agents trained on unified data streams.

“Before, they were reacting to problems. Now, they’re anticipating them,” he said. “And they didn’t need new hardware — just a smarter way to use the data they already had.”

He also emphasised that this architecture is scalable for mid-sized manufacturers: “You don’t need a data centre the size of a football pitch. You need a data strategy that understands time.”

Strategic takeaways
1. Build AI on a time-aligned foundation
AI models thrive on clean, contextualised data. The unified time space ensures every data point — from ovens to ERP — is indexed in time, enabling AI to detect patterns, predict outcomes, and optimise processes across the entire production ecosystem.

2. Accelerate cross-plant intelligence
By harmonising data across lines, sites, and systems, manufacturers can deploy AI models that scale beyond pilot projects. This unlocks enterprise-wide visibility and enables centralised optimisation of energy, throughput, and quality.

3. Future-proof traceability and compliance
Unified time space simplifies audit trails and regulatory reporting. With every event timestamped and contextually linked, manufacturers gain robust traceability — from raw material intake to finished product dispatch.

4. Enable predictive and adaptive operations
Real-time, time-aligned data empowers AI to move from reactive alerts to predictive insights and adaptive control. This means fewer unplanned downtimes, smarter resource allocation, and dynamic scheduling.

5. Democratise data access for all functions
Whether it’s quality assurance, maintenance, or sustainability teams, a unified data layer ensures everyone works from the same source of truth. This fosters cross-functional collaboration and speeds up decision-making.

6. Scale without rebuilding infrastructure
Torralba emphasised that manufacturers don’t need new hardware — just smarter architecture. The unified time space overlays existing systems, making it a cost-effective path to AI scalability.

The unified time space isn’t just a backend upgrade — it’s a front-line enabler of digital transformation, sustainability, and competitive agility.

“AI is not a magic wand,” Torralba concluded. “It’s a tool. And like any tool, it needs the right foundation. The unified time space is that foundation.”