The Real Battle in Enterprise AI: Owning the Integration Layer
The AI integration layer is the set of data, content, workflow and connectivity capabilities that allow AI agents to understand enterprise context and take governed actions across many systems in a consistent, automated way.
The key story in enterprise software M&A is no longer about buying customer-facing apps; it is about buying the plumbing that AI needs to operate. Salesforce’s acquisition of Informatica for USD 8 billion (approx. RM37.0 billion) in 2025 strengthened its data integration, governance and metadata management foundations for enterprise AI deployment. Its newly announced USD 3.6 billion (approx. RM16.7 billion) purchase of Fin pushes that strategy into autonomous customer service, with agents operating across chat, email, SMS and voice support. Meanwhile, Genesys has acquired Pinkfish, adding Model Context Protocol (MCP) tool integrations and agentic workflow automation into its contact center AI stack. These moves are not side bets — they are a deliberate land grab for the AI integration layer. Vendors know that whoever controls the connective tissue will control how AI agents behave, what data they see and which workflows they can automate.
Salesforce Is Building an AI Operating System, Not Just a Bigger CRM
Salesforce is assembling an enterprise AI operating layer that connects data governance, workflow automation, collaboration and autonomous customer service agents, rather than merely expanding its CRM footprint.
Look at the pattern: MuleSoft connects applications and data; Tableau gives analytics; Slack provides collaboration; Informatica adds governed, AI-ready data; Contentful supplies the content layer; and Fin delivers autonomous agents. MuleSoft connects agents to other systems, Informatica helps them trust the data they act on, Tableau helps humans understand what agents did, while Slack lets humans and agents collaborate. Contentful’s content APIs plug into Salesforce’s Data 360 and Agentforce so AI agents can assemble and deliver personalized content dynamically across channels instead of relying on static pages. One quotable takeaway is this: “Salesforce is assembling an enterprise AI operating layer — not just expanding CRM — connecting data governance, workflow automation, collaboration and autonomous customer service agents.” This is not incremental CRM; it is a bid to become the default AI coordination layer for enterprise customer experience.
Genesys and Pinkfish: Contact Center AI as an Agentic Orchestration Hub
In contact center AI, the integration layer is where the real power shift happens: it is where customer intent gets translated into reliable, auditable actions across CRM, ERP and other back-office systems.
Genesys, a cloud-based customer experience and contact center platform provider, announced on June 30 that it acquired Pinkfish, an agentic orchestration workflow company. The deal adds MCP-based tool integration and workflow automation to Genesys Cloud AI, bringing more than 500 integrations and 25,000 MCP tools covering CRM, ERP, IT, HR, order management and billing applications. Pinkfish effectively turns Genesys into an agentic workflow automation hub: AI agents can call MCP tools, maintain state across systems and complete end-to-end work instead of handing tasks off to humans midway. The company expects Pinkfish capabilities to reach Genesys Cloud customers through its marketplace by the end of July, with native integration planned by the end of its fiscal year on January 31, 2027. For enterprises, this means contact centers become orchestration brains, not just service channels.
Why Vendors Are Buying, Not Building, Their AI Integration Stacks
Enterprise software vendors are choosing acquisition over in-house development because the integration and workflow automation problem is too sprawling, and AI adoption timelines are too short, to start from scratch.
Salesforce has spent several years expanding beyond CRM through acquisitions in customer experience, data, AI and digital engagement rather than building every component internally. Its deals for MuleSoft, Tableau, Slack, Informatica, Contentful and Fin show a clear preference for buying proven components of the AI integration layer and then trying to unify them. In parallel, Genesys is adding Pinkfish’s 25,000 MCP tool integrations to connect AI-driven workflows to existing CRM, ERP and back-office systems instead of developing hundreds of connectors on its own. This is part of a broader enterprise software M&A wave: vendors are consolidating integration and agentic workflow automation capabilities to accelerate AI adoption. The logic is simple but blunt: whoever assembles the widest, most reliable AI integration layer fastest will become the default platform that everyone else plugs into.
Control of Legacy Connections Is the Next Lock-In — and the Next Risk
These deals position acquirers to control how enterprise AI agents connect to legacy systems and data sources, for better and for worse.
By owning the integration and AI operating layers, vendors like Salesforce and Genesys decide which systems are first-class citizens in AI workflows and how governance is enforced. MuleSoft, Informatica, Tableau and Slack each play a role in how agents see, trust, explain and collaborate around enterprise actions. Pinkfish’s 500+ integrations and 25,000 MCP tools give Genesys a central role in connecting customer intent to governed actions across CRM, ERP, IT, HR and billing platforms. According to one source, “the autonomous enterprise depends on the ability to coordinate actions across complex business environments while maintaining governance and control.” The upside is clear: organizations that deployed autonomous AI systems reported a 28% improvement in issue resolution time and a 19% increase in first-contact resolution rates. The risk is equally clear: if one vendor owns your AI integration layer, switching costs and dependency rise sharply. The conclusion is unavoidable: enterprises should treat AI integration strategy as a control point, not a footnote.
