Enterprise AI Agents Don't Need Platform Overhaul (2026)

Thesis: The Agent Context War Isn’t About Migration—It’s About Preservation

Every enterprise AI vendor now agrees that context separates working agents from expensive demos. The disagreement is over how you deliver it.

Hyland CEO Jitesh Ghai is betting against the dominant playbook. While competitors push cloud migration, process redesign, and infrastructure overhauls to make enterprises “agent-ready,” Ghai argues the opposite: the fastest path to useful AI agents runs through your existing systems, not around them.

At CommunityLIVE 2026 this week, Hyland doubled down on this thesis with general availability of its Enterprise Context Engine and Enterprise Agent Mesh, plus a new headless mode that turns its content platform into consumable infrastructure. The message is clear—context comes from meeting organizations where they are, not where vendors wish they’d be.

This isn’t just product positioning. It’s a structural argument about how regulated industries will actually adopt agentic AI, and it runs counter to nearly every enterprise software vendor’s current strategy.

Evidence: Why the “Blow It Up” Playbook Fails in Regulated Industries

The vendor consensus strategy looks like this: move data to cloud, redesign business processes, implement enterprise-wide change management, then deploy agents. Ghai calls this “blowing things up” and considers it both unnecessary and improper for Hyland’s customer base.

The numbers explain why. Hyland operates at over $1 billion in revenue with 15,000 customers concentrated in healthcare, insurance, banking, and government. These aren’t startups that can replatform on a quarter’s notice. They’re organizations where unstructured documents are central to operations and compliance requirements make architectural decisions multi-year commitments.

Ghai estimates knowledge workers in these sectors spend 20-40% of their time on what he calls “human ETL”—manually extracting, transforming, and loading information from documents. He pegs 70-90% of enterprise data as unstructured, with most sitting in content management systems.

The technical bet is that LLMs finally make this content tractable without migration. Hyland’s Enterprise Context Engine layers AI structuring on top of existing content repositories, building knowledge graphs enriched with industry-specific ontologies for healthcare, insurance, financial services, education, and government.

This ontology layer is where Ghai thinks most vendors underestimate complexity. “You and me playing with Claude Code, we have our context window, and we can throw a bunch of stuff in there and get all sorts of value,” he told The New Stack. “At enterprise scale, it’s a whole other exercise.”

The architecture reflects this: Content Innovation Cloud federates existing systems, AI structures unstructured documents, knowledge graphs add business context, and the Enterprise Context Engine governs the entire layer. Agents consume this context through APIs—they don’t require rearchitecting the systems that generated it.

Hyland’s Agent Lifecycle Management framework adds the governance layer regulated industries demand: an Agent Library cataloging every agent, base agent archetypes, and an “Agent Passport” that defines identity, capabilities, guardrails, and compliance status before production deployment. The upcoming Control Tower provides observability into agent decision pathways.

Context: The Enterprise Content Management War Becomes an Agent Context War

Hyland isn’t alone in positioning content management as the context layer for enterprise AI. OpenText frames Content Cloud as agent context infrastructure. Box pushes to become the enterprise content hub. The entire category now treats context as the competitive moat.

This shift reflects a broader pattern in enterprise software. As AI agents move from prototype to production, the bottleneck isn’t model capability—it’s organizational context. Generic foundation models know how to reason, but they don’t know your company’s policies, your customer history, your regulatory requirements, or which document in which system contains the answer to a compliance question.

The traditional enterprise playbook for new technology—consolidate, standardize, migrate—makes sense when you’re building greenfield systems. It falls apart when the systems you need context from are the same systems running mission-critical operations in regulated industries.

Ghai’s background adds credibility to this thesis. He spent years as chief product officer at Informatica, bringing a structured-data perspective to Hyland’s unstructured content core. The synthesis of both worlds—structured data from third-party systems plus structured views of unstructured content—is what Hyland calls its “content and data fabric.”

The headless mode announcement signals where this goes next. By exposing the Context Engine as consumable APIs, Hyland moves from packaged applications to infrastructure. Data engineering teams, ISVs, and platform ecosystems like Databricks and Snowflake can pull Hyland’s enrichment and governance without adopting its interface.

This is a bet on fragmentation as the steady state. “There’s going to be fragmentation, and we recognize the value we uniquely deliver, which is context from your content and data and processes,” Ghai says. “We recognize that there are other vendors that could equally benefit; their agents could equally benefit from this. And that is why we believe we have to be independent and neutral, open and modular.”

Neutrality becomes strategic when no single vendor will own the entire agent stack. If enterprises end up running agents from multiple sources—vendor-built, custom-developed, third-party—the context layer that serves all of them wins.

Counterarguments: When Preservation Becomes Technical Debt

The strongest counterargument to Ghai’s thesis is that “meeting organizations where they are” often means inheriting decades of technical debt. Legacy content management systems weren’t designed for AI consumption. They’re siloed, inconsistently tagged, poorly governed, and often running on-premises with limited API access.

Federation sounds elegant until you’re federating 15 different content repositories with incompatible metadata schemas. Building a unified knowledge graph across that fragmentation requires either forcing standardization (which contradicts the “don’t blow it up” promise) or accepting a lowest-common-denominator context layer.

The cloud migration playbook exists for a reason: it consolidates, it modernizes APIs, it enables the kind of elastic compute AI workloads demand. Vendors pushing rearchitecture aren’t wrong about the technical benefits—they may just be wrong about regulated industries’ willingness to pay the switching cost.

There’s also the question of whether Hyland’s ontology approach scales beyond its core verticals. Industry-specific knowledge graphs work when you’re deep in healthcare or insurance workflows. They’re harder to build for horizontal use cases or companies that span multiple sectors. The ontology that makes context useful in one domain may not transfer.

The governance layer could become a bottleneck. Agent Passports and Control Towers sound responsible, but they add friction. If every agent requires certification before deployment, experimentation slows. The companies that win on agents may be the ones willing to run faster with less governance, at least in non-regulated domains.

Still, Ghai’s bet is specifically on regulated industries where governance isn’t optional and migration timelines stretch across years. For that customer base, the counterarguments carry less weight.

Predictions: What Happens When Context Becomes Commoditized Infrastructure

By Q4 2026, at least two major enterprise software vendors will announce similar “agent context without migration” offerings in direct response to Hyland’s positioning. The playbook is too obvious to ignore: federation layer, AI structuring, industry ontologies, governance framework, headless APIs. Expect OpenText and Box to make aggressive moves here.

By mid-2027, we’ll see the first high-profile agent deployment failure blamed on inadequate context infrastructure. Some Fortune 500 company will launch customer-facing agents without proper grounding in enterprise systems, produce hallucinated responses in a regulated context, and trigger either regulatory action or a PR crisis. This will validate the “context is the moat” thesis and accelerate enterprise spending on context layers.

The real test for Hyland’s approach comes in 12-18 months when enterprises try to scale from pilot agents to hundreds of agents. If federation and ontology can deliver coherent context across that complexity, the “don’t blow it up” thesis wins. If organizations hit a ceiling where they need migration to scale, the traditional playbook rebounds.

Headless mode will be Hyland’s fastest-growing revenue stream by 2028. As agent fragmentation increases, infrastructure that serves multiple agent platforms becomes more valuable than any single agent platform. Content management vendors who successfully pivot to infrastructure will outperform those who stay in applications.

The meta-prediction: context layers will consolidate faster than agent platforms. We’ll see five major enterprise context engines by 2027—Hyland, OpenText, Box, plus two from either Microsoft, Oracle, or Salesforce. Meanwhile, the agent platform layer will remain fragmented across dozens of vendors. The inverse of the application layer pattern.

Ghai’s thesis isn’t that agents don’t need context—it’s that context doesn’t need migration. If he’s right, the fastest path to enterprise agents runs through preservation, not disruption. The next 18 months will tell us whether regulated industries agree.