The Enterprise Agent Platform Land Grab: Why Incumbents With Context Advantage Will Win the "Human-Agent Team" Race

A popular thesis in venture circles holds that AI agents will hollow out enterprise SaaS. The argument goes like this: agents will abstract away the interface layer, users will interact with models rather than applications, and the trillion-dollar SaaS ecosystem will be reduced to plumbing behind an API. Salesforce becomes a dumb database. Asana becomes a task log. The agent becomes the platform.

This thesis has it precisely backwards.

This week offered evidence to the contrary. On May 28, Asana acquired StackAI for $75 million , explicitly positioning itself as "the operating system for human-agent teams." On the same day, Glean announced it had crossed $300 million in annual recurring revenue — tripling its top line in 15 months — by selling major enterprises — including Samsung, Pinterest, and Databricks — on AI budget-cutting through its enterprise context graph. These are not signals of disruption. They are signals of absorption.

Incumbent enterprise platforms are not being replaced by agents. They are embedding agents into their existing data gravity. For CTOs making platform decisions today, that difference is everything.

The Asana Signal: Workflow as the Operating System

Asana’s acquisition of StackAI is the clearest example yet of the integration-over-rip-and-replace strategy. StackAI had raised under $20 million before the acquisition. It built a no-code agent builder that connected systems like Salesforce, Slack, and Google Workspace. On its own, it competed with Zapier, OpenAI, and Anthropic — an increasingly crowded space with thinning margins.

Inside Asana, StackAI slots in as the execution engine for AI Teammates , Asana’s pre-built agent roster: the Campaign Brief Writer that reads project notes and Google Docs to produce structured briefs, the Brand Auditor that reviews creative against guidelines, the Launch Planner that maps dependencies across teams and timelines.

These agents don’t need to discover your organizational structure or guess your permissions. They inherit both from Asana’s existing workspace. They know which projects you own, which documents you can read, which approvals you need. The governance layer is already built — not bolted on after the fact.

Asana CEO Dan Rogers framed it directly: “This acquisition accelerates our roadmap and takes us into the next phase of human-agent work.” The key word is accelerates, not invents. Asana already owned the data model. StackAI gave it the orchestration layer to execute across systems.

The pattern is not “agents replace Asana.” The pattern is “Asana becomes the agent platform.”

The Glean Signal: Context Is the Moat

Glean’s $300 million ARR milestone matters even more for leaders making infrastructure decisions. Glean started as enterprise search — a better way to find files across Slack, Jira, Salesforce, and Google Drive. But the company’s strategy shifted from “find information” to “ground AI in your specific business context.”

CEO Arvind Jain described the shift: “The AI models themselves don’t really understand anything about your business […] so you have to connect the reasoning and generative power of the models with the context inside your company.” Glean’s answer is the context graph — a persistent map of how information, people, permissions, and decisions connect across an enterprise.

What makes this defensible isn’t the technology. It’s the accumulated data. The connectors Glean has built to Salesforce, the permissions it has indexed, the search patterns it has observed — all of these build switching costs that compound over time. Any competitor can call the same LLM API. No competitor can replicate seven years of enterprise indexing relationships without starting from scratch.

Glean’s other selling point gets at the real dynamic here: it sells on AI budget reduction. Jain claims Glean reduces token consumption by giving agents the exact information they need rather than requiring them to search blindly. In a world where enterprise AI spend is growing faster than any other line item on the P&L, saving tokens through a better context layer is a CTO’s dream pitch.

The Counter-Narrative That Venture Capital Misses

The dominant venture thesis holds that startups building “agent operating systems” will win because incumbents are too slow and siloed, too invested in the old paradigm. Foundation Capital’s recent essay on context graphs argues that “systems of agents” startups have a structural advantage because existing SaaS platforms capture current state, not decision traces. The essay also warns that agents built by incumbents “inherit their parent’s architectural limitations.”

This argument is strongest against record systems like Salesforce and Workday. Salesforce’s Agentforce inherits the architectural reality that CRM stores current state, not the decision path that led there. When a discount gets approved, Salesforce records the final price but not the context — the SEV-1 incidents, the VP’s Slack approval, the precedent from last quarter.

The argument is weakest against workflow systems like Asana, and knowledge platforms like Glean. These platforms sit in the execution path. They see the work as it happens. Asana knows which tasks were reprioritized and why. Glean indexes the documents and conversations that contain the actual reasoning. These are not record systems. They are trace systems — and they are structurally positioned to capture the decision traces that Foundation Capital correctly identifies as the next trillion-dollar asset class.

The Three Types of Incumbent, and Which One Wins

For CTOs evaluating where to place their bets, the enterprise agent platform landscape is resolving into three categories:

• Record platforms (Salesforce, Workday, SAP) own canonical data about customers, employees, and operations. Their agent strategy is constrained by the fact that they capture what happened, not why. They will remain essential but will likely not become the agent operating system.
• Knowledge platforms (Glean, Microsoft Viva) own the context graph — the map of information, people, and permissions. They are structurally positioned to become the intelligence layer that routes information to agents efficiently. Their weakness is that they sit in the read path, not the write path. They see your documents but don’t execute your workflows.
• Workflow platforms (Asana, ServiceNow, Jira) own the execution path — the actual orchestration of tasks, approvals, and handoffs across teams. They see the decision as it happens. They know who approved what, when exceptions were granted, and how work actually flows. Their weakness is breadth: they need to absorb or integrate with the knowledge and record layers to provide complete context.

The workflow platform that integrates deepest with your knowledge and record layers will become your de facto human-agent operating system.

Here’s the thesis I’d put in front of any engineering leadership team: the workflow platform that integrates deepest with your knowledge and record layers will become your de facto human-agent operating system. That’s what Asana’s StackAI acquisition is designed to achieve. It’s also what ServiceNow’s Now Assist is doing, and what Atlassian’s Rovo is reaching for.

What Leaders Should Do Now

Expect a wave of acquisitions in this space over the next 12 months. Every incumbent with a workflow graph and a permission model will look to acquire agent orchestration capability. Every company sitting on a knowledge graph will need to decide whether to build or buy the execution layer.

For CTOs, the practical takeaway:

1. Do not build your own agent middleware. The abstraction layer between models and enterprise data is becoming a commodity war that startups are losing to incumbents with deeper integrations. The Gleans and Asanas of the world are investing billions to own this layer. You cannot out-build them, and you should not try.
2. Bet on the platform that already knows how your organization works. The switching cost isn’t the agent’s model — it’s the accumulated context. Each indexed document, mapped permission, and traced workflow makes it harder to leave. Choose a platform that already sits inside your existing processes rather than one that requires you to rebuild them.
3. Watch for the “integration tax.” The biggest risk to incumbents is not that startups build better agents — it’s that incumbents lock in their agent platforms through data egress fees, restrictive APIs, and walled-garden connectors. Foundation Capital flags this correctly: “They’ll lock down APIs and adopt egress fees to make data extraction expensive.” A CTO’s negotiating leverage is highest right now, before the platforms mature.

The race to own human-agent teams is not a technology race. It’s a data advantage race. The winner will not have the best model. The winner will have the deepest understanding of how your company actually works.

For most enterprises, that understanding already lives inside your existing SaaS stack. The question is whether your vendors turn it into an operating system — or take it hostage.

Further Reading

1. “Long Live Systems of Record” — Jamin Ball / Clouded Judgement — The piece that kicked off the “agents vs. systems of record” debate, arguing agents raise the bar for systems of record rather than replacing them.
2. “AI’s trillion-dollar opportunity: Context graphs” — Foundation Capital — Argues that capturing decision traces — not just data snapshots — will create the next generation of trillion-dollar platforms, and that incumbents are structurally disadvantaged.
3. “The enterprise AI land grab is on: Glean is building the layer beneath the interface” — TechCrunch — February 2026 interview with Glean CEO Arvind Jain explaining the shift from search to middleware layer, and why he sees OpenAI and Anthropic as partners, not competitors.
4. “Asana acquires no-code agent-builder StackAI” — TechCrunch — Details of the $75M acquisition and Asana’s positioning as “the operating system for human-agent teams.”
5. “Glean’s top line crosses $300M” — TechCrunch — Analysis of Glean’s revenue milestone, its consumption-based pricing model, and the AI budget-cutting narrative driving enterprise adoption.