The Headlines vs. The Undercurrents

Salesforce's is acquiring Informatica for $8 billion. The official narrative? To "enhance Salesforce's trusted data foundation critical for deploying powerful and responsible agentic AI." A sensible, forward-looking statement. Yet, as we at Activant Research dig deeper, we see this move as far more than a simple AI enablement play. It's a potent signal of a fundamental disruption Salesforce is grappling with: the seismic shift from Systems of Record (SoR) to Systems of Intelligence (SoI). This isn't just another enterprise software acquisition. It's an $8 billion alarm bell signaling that we've reached an inflection point where the fundamental value proposition of enterprise software is shifting from capturing data to acting on it autonomously.

We've emphasized that data is the undervalued element of AI. While algorithms and models capture headlines, their efficacy in the real world is entirely dependent on the quality, accessibility, and context of the data they act on.

Salesforce's own massive push towards its AI Agent product, "Agentforce," underscores this. With an impressive 3,000 customer sign-ups, concrete case studies of value creation are sparse. Consider their Q4 FY25 earnings call, where Salesforce called out two internal case studies and only one external - that "For OpenTable, Agentforce handles 73% of all restaurant web queries". This is a good stat but feels like it leaves something lacking, particularly given the power of AI Agents and the potential across a scale of 3,000 customers.

If Salesforce is facing a challenge in deriving value from their AI Agent products, their splurge on Informatica might reveal it. As we flagged in September 2024: 70% of organizations find their data to be a significant hurdle in capturing value from Generative AI tools, citing it as their most common issue.

Agentic AI, the kind Agentforce aspires to be, doesn't just need data; it needs a sophisticated data ecosystem. These agents require:

• Rich Metadata: To provide a contextual understanding of ingested data. For example, a customer support AI Agent responding to a support ticket will reference critical metadata including the ticket status and information on the requester.
• Data Lineage: For explainability of AI actions, crucial for regulation and compliance. Audit teams will require that an AI system producing financial reports can demonstrate exactly which, timestamped, version of source data was used to produce a report.
• Robust Data Governance: To ensure agents are querying and acting upon accurate, updated and permissioned data. An AI chatbot that HR departments provide to employees for quick answers to questions on company policies should not reveal the CEO’s salary.

This is where Informatica, a long-standing leader (the Gartner kind) in data integration, quality, and governance, ostensibly fits in. The acquisition appears to be a direct attempt to bolt on these critical data layer capabilities and start to create real value for Salesforce’s Agentforce customers.

However, at Activant we see something deeper in the deal.

The System of Intelligence Gambit

While AI agents are the immediate catalyst, the Informatica acquisition speaks to a more profound architectural evolution: the transition from Systems of Record to Systems of Intelligence.

• Systems of Record (SoR): These are the foundational databases and applications (like Salesforce's core CRM) that capture and store enterprise data. In the SoR era, value was derived from efficiently organizing and providing human access to this data. They were the digital filing cabinets, a massive productivity boost over spreadsheets and paper.
• Systems of Intelligence (SoI): These are emerging layers that sit atop SoRs, designed not just to store data, but to actively leverage it. SoIs apply analytics, machine learning, and now, agentic AI, to drive automated actions, generate insights, and unlock new value propositions. Where SoRs empowered humans, SoIs promise to automate vast swathes of human labor, creating and capturing exponentially more value.

This shift poses an existential question for SoR giants like Salesforce: do they remain the underlying data repository while new SoI players capture the lion's share of the value, or do they fight their way "up the stack" to become an SoI themselves?

Well, Salesforce are acquiring Informatica and ServiceNow acquired data catalog and metadata company data.world (to launch their "workflow data fabric"). These are the exact tools needed to build out the data control layer that AI products will need. The SoRs are racing to build out the capabilities to not just hold data, but to intelligently act on it.

The risk is palpable: over a long enough time horizon, SoIs that consistently pull data from SoRs, enrich it, and drive autonomous actions could, in effect, become the new SoR, or at least diminish the strategic importance of the original data engines.

The Iceberg Effect

Underpinning this SoR-to-SoI transition is a quieter, but equally transformative, revolution in data architecture - the rise of open table formats like Apache Iceberg.When Databricksacquired Tabular (the company behind Iceberg) and Snowflake open-sourced its Polaris Iceberg catalog, they weren't just making technical announcements, they were declaring that the era of data lock-in is over.

With the market standardizing on Iceberg, it becomes possible to perform far simpler queries to external data sources, which come with rich metadata and the ability to do “no-copy” data federation. In short, it is getting much easier to move data between systems. The era of data moats is ending and its possible that soon, printing billions of dollars as a database with great UI and some workflow functions will no longer be an option.

So, the Salesforce-Informatica deal is a bold move in a high-stakes game. It’s less about enhancing a single product (Agentforce) and more about repositioning Salesforce for an era where their biggest moat, data, is eroding. The new moats are intelligence.
The winners in this next wave will be those who can:

1. Master the complexities of enterprise data with robust metadata, lineage, and governance.
2. Embrace open standards and interoperability to leverage data wherever it resides.
3. Successfully build and deploy Systems of Intelligence that deliver tangible value.

The Innovator's Dilemma in Action

We're watching the innovator's dilemma play out in real-time. Incumbents like Salesforce and ServiceNow are trying to transform their businesses. History suggests this kind of transformation is nearly impossible for incumbents to execute and Clayton Christensen's research shows us that disruptive innovations typically come from new entrants who can build from first principles without the burden of legacy architectures.

And that gets to one of the core issues at the heart of this competition – are we really to believe that legacy platforms, who themselves are a patchwork of platforms acquired over the years, will be the best place to unify and query enterprise data? Are these visionary moves to build impenetrable, end-to-end platforms? Or are they reactive scrambles to plug critical gaps in clunky, legacy, and fragmented architectures, as the AI wave exposes existing weaknesses?

This acquisition is a clear indicator: the race to build, and own, the enterprise system of intelligence is well and truly on. And data, in all its complex glory, is the battlefield.