Enterprise AI Has a Massive Problem, and It’s Not the Models

As enterprises accelerate the deployment of AI, discussions are shifting from models to data. At Informatica World 2026 in Las Vegas last week, the company repeatedly highlighted that “trusted enterprise context,” rather than just raw data, will determine how well organisations can scale AI systems in production.

The event focused heavily on agentic AI, metadata intelligence, governance, and what Informatica described as the industry’s first “headless data management platform”. 

The company also highlighted deeper integrations with Salesforce following the acquisition of Informatica, positioning the partnership as a foundation for enterprise AI workflows spanning customer engagement, governance, and automation.

While the messaging at the summit leaned strongly towards the future of autonomous AI systems, customer conversations at the event reflected a more grounded reality: enterprises are still focused on building reliable data foundations before AI can scale meaningfully.

One such example is “Yum! Brands”, an American multinational fast food chain, which has been working with Informatica for over a decade to manage data pipelines and governance across its global restaurant ecosystem.

From Data Pipelines to AI Readiness

During a conversation with AIM, Kartik Pillai, Director of Data Strategy, Master Data Management, AI & Data Governance at Yum! Brands, says the company’s Informatica journey initially began as a traditional ETL and warehouse modernisation effort.

“We started with Informatica primarily as tooling for our ETLs to manage the data pipelines going from our sources into our data warehouse,” Pillai says. “But about 15 to 18 months ago, when the whole world pivoted towards AI and ML, we introspected and asked ourselves if we were ready for AI.”

That exercise, he says, exposed gaps in governance, master data management, and data quality.

Yum! Brands, which operates over 63,000 restaurants across 155 countries and territories, now uses Informatica’s Intelligent Data Management Cloud (IDMC), master data management (MDM), data governance, change data capture, and data quality tools as part of a broader AI readiness strategy.

Its first large-scale MDM deployment, Store MDM, has already been running in production for over a year.

The complexity of that transformation is amplified by Yum!’s operating structure. Nearly all of its business runs through franchise models, generating large volumes of operational, customer, and logistics data across regions and brands.

Pillai says one of the biggest reasons Yum! deepened its investment in Informatica was the need for a system capable of operating across a “multi-hyperscaler, multi-vendor ecosystem”.

“We needed someone who could play the middle fiddle where every hyperscaler, every brand, and every ecosystem is taken care of at the scale we operate,” he says.

Governance Moves to the Centre of AI

A recurring theme at Informatica World was that enterprises do not simply have a “data problem” but a “context problem”. The leadership argued that AI systems without trusted context could generate inaccurate decisions, poor customer experiences, and governance failures.

That concern is increasingly resonating with enterprises deploying AI in production environments.

According to Pillai, governance at Yum! has evolved from a reporting and compliance exercise into a foundational layer for enterprise-wide AI systems.

“Governance and data quality are not an afterthought anymore. They are becoming part of the delivery process itself,” he says. They are to build that contextual layer as they start deploying AI on top of it,” he says. 

Pillai acknowledges that governance is often perceived internally as something that slows organisations down. However, AI adoption has helped shift that perception.

“People who have deployed AI products in production quickly realise that without governance and data quality, it is not scalable,” he says.

That perspective aligns with Informatica’s message at the conference, where the company highlighted metadata, lineage, and governance as essential infrastructure for agentic AI systems. The slides from the event referred to “trusted enterprise context” as “the new currency of AI”.

The company also announced new CLAIRE AI agents focused on data discovery, data quality, metadata enrichment, stewardship, and data engineering automation.

AI Personalisation Meets Regional Complexity

Like many quick-service restaurant companies, Yum! is increasingly deploying AI systems to personalise customer engagement and streamline restaurant operations.

Pillai says customer-facing systems now focus heavily on recommendation engines, loyalty experiences, and reducing “decision fatigue” within digital ordering journeys.

“Taco Bell’s app experience is really good because it is personalised,” he says. “If you are a loyalty member, the app recommends what you should get, and that removes the fatigue customers have to make decisions.”

The company is also deploying internal AI systems for restaurant employees, including chatbots that assist frontline workers with troubleshooting operational issues and daily workflows.

Importantly, Pillai says the company does not view AI as a headcount reduction exercise.

“We deploy AI workflows to elevate employees to do more value-added work rather than monotonous tasks,” he says.

However, scaling these systems globally introduces significant regional complexity, especially around localisation, consent frameworks, and customer behaviour.

Pillai notes that Southeast Asian markets differ substantially from Western markets in areas such as mobile ordering, QR-based commerce, WhatsApp ordering, and loyalty adoption.

Those differences also affect how data governance and customer engagement systems are implemented.

Yum! maintains regional data warehouses to comply with localisation requirements and avoid cross-border movement of personally identifiable information (PII). According to Pillai, customer data remains within regional boundaries unless anonymised.

Beyond the AI Hype Cycle

While Informatica World 2026 strongly emphasised agentic AI and autonomous workflows, enterprise customers at the event appeared more cautious about how quickly those systems can be deployed in real-world operations.

Yum!’s approach reflects that caution.

Pillai says predictive AI models already help forecast staffing needs, delivery times, and restaurant demand, but human oversight remains essential, particularly during volatile economic or geopolitical conditions.

“The models are as accurate as the data you feed them,” he says. “But we still believe human intuition and experience are valuable, especially for restaurant managers who have spent decades operating restaurants.”

That balance between automation and human decision-making may increasingly define how enterprises approach AI over the next few years.

For companies such as Yum! Brands, the immediate priority appears less about replacing human systems entirely and more about building reliable infrastructure capable of supporting AI at scale without losing operational control.