Workday just acquired Pipedream, an integration platform with 3,000+ pre-built connectors. That's their sixth AI-related acquisition since February:
- HiredScore
- Evisort
- Sana Labs for $1.1 billion
- Flowise
- And now Pipedream
Six infrastructure acquisitions in under a year—Workday isn't building agent features, they're assembling control points for where agents can operate across enterprise software.
Look past the agent features and reasoning capabilities. Workday is assembling the infrastructure layer that determines where agents can operate reliably across enterprise software. That's control point acquisition.
Enterprise Software Is Becoming Something Else
Enterprise software platforms are becoming something fundamentally different. Workday isn't just adding AI agents to their HR and finance products. They're positioning to be the infrastructure that makes everyone else's agents work across enterprise systems.
The Pipedream acquisition shows why. Pipedream's managed authentication handles OAuth flows, token refresh, and delegated authority across 3,000+ applications. This is the connection layer. The infrastructure that determines whether an agent can actually execute tasks across the fragmented landscape of enterprise software.
If you control the connection layer, you control where agents can operate. Through operational reality. Authentication across thousands of applications—each with regional variations, undocumented rate limits, permission models that change quarterly—requires infrastructure that only exists in a few places.
One company spent over $1 million building 250 integrations over two years. Pipedream offered 2,500 immediately.
The companies that built this infrastructure early become strategic chokepoints. The window for building equivalent breadth has effectively closed.
Why Six Acquisitions in Under a Year Matters
Six acquisitions in under a year isn't normal for enterprise software. CEO Carl Eschenbach was direct:
"We can't all do it ourselves."
When Gartner predicts 40% of enterprise applications will feature AI agents by 2026—up from less than 5% today—that's a land grab for the infrastructure layer before the market crystallizes around a few platforms. The pace matters.
The companies assembling complete agent infrastructure stacks now are drawing the boundaries of what becomes possible for everyone else. If Workday succeeds in becoming the connection layer for enterprise agents, what leverage does that create? What happens to independent agent builders who need those connections to operate?
If This Signal Holds
The agent market won't be won by the most sophisticated reasoning engines or the most elegant user interfaces. It'll be won by whoever controls the infrastructure layer that determines where agents can operate reliably at scale.
Your enterprise software vendor isn't just providing applications anymore. They're potentially becoming your agent infrastructure provider. Different lock-in dynamics. Different strategic dependencies.
For companies building agents, the infrastructure layer is consolidating faster than the typical enterprise software cycle. The platforms that can deploy agents reliably across thousands of third-party systems now will capture the growth wave. The ones still building integration infrastructure will find themselves operating in an ecosystem where the connection points are already controlled.
Watch where the infrastructure layer consolidates. That's where the strategic dependencies form.
Things to follow up on...
-
Pipedream's agent-specific products: The company launched String.com and Pipedream Connect in 2025 specifically designed for building and deploying AI agents with natural language interfaces and managed authentication.
-
The delegated authority challenge: AI agents face a critical security problem called "brokered credentials" where LLMs must act on behalf of users without ever seeing actual API keys or tokens.
-
Workday's integration timeline: The company plans to combine Pipedream, Sana, and Flowise capabilities during 2026, possibly in the first half, bringing three acquisitions onto the Workday AI platform in quick succession.
-
OAuth limitations for agents: Traditional authentication protocols were designed for human users with browsers, making them awkward for headless agents that need dynamic, context-aware permissions rather than static scopes.

