The defensible asset in enterprise AI is not the model. It’s the organizational world model.

Every major compute era decomposes into specialized layers with different winners at each level. Cloud split into IaaS, PaaS, and SaaS. The modern data stack split into ingestion, warehousing, transformation, and BI. Each time, specialists beat the generalists because each layer has its own economics: its own rate of change, capital requirements, and sources of lock-in.

The enterprise AI agent stack is doing the same thing right now. Arvind Jain, the CEO of Glean, recently published a structural analysis of the emerging enterprise agent architecture that crystallized something I’d been thinking about. He describes a stack splitting into six layers (security, context, models, orchestration, agents, and interfaces), and each layer is defensible to a different degree. Glean sits in the context layer so the usual positioning caveats apply, but the structural argument is sound regardless of who makes it.

The three sections below cover models converging toward shared infrastructure, the two depths of the context layer, and the compounding flywheel that context and orchestration form.

Enterprise AI agent stack diagram showing six layers ranked by defensibility: Context scores highest (hardest to rebuild), followed by Orchestration and Security, while Models and Interfaces have the lowest switching costs

I. Models are converging toward shared infrastructure

The model layer is converging fastest toward commodity economics. Training costs scale roughly 2.4x per year, with current frontier runs costing hundreds of millions and billion-dollar training runs already underway, according to Anthropic’s Dario Amodei. Only a handful of organizations on Earth can operate at this scale: OpenAI, Google DeepMind, Anthropic, Meta, and a few others including xAI and Mistral. This is capital-intensive infrastructure, with the same structure as semiconductor fabs or cloud hyperscalers. I expect foundation models to become shared utilities, so no enterprise will build a moat on one.

Companies already act on this. 37% of enterprises now use five or more models in production, up from 29% the prior year. Different tasks demand different models: Claude for code and tool use, GPT for extended reasoning, Gemini Flash for low-latency routing, specialized models for image generation and embeddings. Betting your enterprise stack on a single model provider carries the same risk as betting it on a single cloud. Open standards like Anthropic’s Model Context Protocol, now hosted by the Linux Foundation with 97 million monthly SDK downloads, and Google’s Agent-to-Agent protocol are making this multi-model enterprise AI architecture practical. If models are infrastructure, differentiation has to come from higher up the stack.

II. The enterprise AI context layer has two depths, and most people only see the first

Most enterprise AI efforts operate at what I’d call Layer 1 context: connecting data sources, indexing content, enforcing permissions, retrieving relevant documents. This is the RAG-era problem set. It is well understood and increasingly commoditized, and virtually every enterprise AI platform offers connectors, vector stores, and retrieval pipelines. Layer 1 matters, but a capability every platform has is hard to defend.

Layer 2 captures process-level understanding. Most enterprise knowledge systems capture decisions, the records that end up in the CRM, the ticketing system, or the ERP. But they don’t capture how those decisions were made: the meetings, Slack threads, document iterations, handoffs, and informal coordination that produced the recorded outcome.

Enterprise AI context layer depth comparison showing what Systems of Record capture (decisions, states, entities, relationships) versus what Context Graphs capture (processes, temporal traces, causal structure, variability), with ML lens annotations mapping to labels versus feature space and trajectory data
In machine learning terms, systems of record give you labels. Context graphs give you the feature space and trajectory data you’d actually need to learn the decision boundary. When your CRM records that Deal X closed at $500K, that’s a label. The context graph captures the features and the trajectory: the 14 meetings, 3 stakeholder handoffs, the pricing negotiation pattern, and the competitive displacement sequence that produced that outcome. An agent trained on labels alone can’t replicate the process that generated them.

This is why so many early enterprise AI deployments produce outputs that are technically plausible but operationally useless. The agent can retrieve the right documents but can’t reconstruct the reasoning process that a human would follow. The highest-value problem in enterprise AI right now is closing that gap with systems that capture and encode process knowledge as well as decision records.

III. Context and orchestration form a compounding flywheel

Reinforcement learning offers an analogy: the orchestrator is the policy, the context graph is the learned world model, and agent traces are the trajectories. Each successful execution reinforces good patterns, and each failure shows where context is missing or stale.

More deployment produces richer traces, which improve the context graph, which improves agent decisions, which builds trust, which drives more deployment.

Organizational world model compounding flywheel showing the five-step loop: Agent Executes → Traces Captured → Context Improves → Better Decisions → More Deployment, with ML analogy mapping table showing enterprise concepts mapped to RL primitives (policy rollout, trajectories, world model update, policy improvement, online learning loop)
This is the same compounding mechanism that makes recommendation engines and autonomous driving systems improve with scale. Netflix gets better at recommendations because every viewing session generates training signal, and Waymo gets better at driving because every mile generates edge cases. In an enterprise, agent traces build an organizational world model, a learned representation of how your specific company works.

Any well-funded lab can approximate model weights. Your organization’s accumulated process knowledge is a moat because it is unique. No one else has your meeting patterns, escalation sequences, or informal decision-making topology.

Where this breaks, and why the agentic AI failure rate will be high

Gartner predicts 40% of enterprise applications will feature task-specific AI agents by 2026, up from less than 5% in 2025. McKinsey’s latest survey shows 23% of organizations are already scaling agentic AI, with another 39% experimenting. But Gartner also warns that over 40% of agentic AI projects will be canceled by end of 2027 due to escalating costs and unclear business value.

The gap between ambition and execution comes from the context problem. Without process knowledge, agents produce plausible outputs that don’t match how the organization actually works. They retrieve the right policy document but apply it without understanding the exceptions your team has developed over years. They draft the right kind of email but miss the relationship dynamics that would change the tone. Shallow context can make even a capable model fail.

Enterprise AI agent stack scatter plot showing six layers plotted by lock-in risk versus rebuild difficulty. Context sits alone in the top-right danger zone with highest lock-in and hardest rebuild. Models, Interfaces, and Agents cluster in the commodity zone at bottom-left. Orchestration and Security occupy the middle.
Models, interfaces, and agents cluster in the commodity zone, where lock-in is low and replacement is easy. Context sits alone in the danger zone, with the highest lock-in risk and the hardest rebuild, so that is where your due diligence should concentrate.

What to actually do about your agentic AI architecture

Don’t go monolithic, because each layer evolves at a different rate. Models improve quarterly, context infrastructure evolves over months, and security requirements shift with regulation. Coupling them into one vendor’s all-in-one platform forces you to upgrade at the speed of the slowest-moving layer. You also inherit the vendor’s architectural bets, integration timeline, and roadmap priorities.

Insist on interoperability: MCP, A2A, and open connectors. If your vendor doesn’t support open standards, you’re absorbing limitations you can’t see yet. You need to be able to swap in new capabilities the moment they appear without rebuilding your stack. The organizations that locked into single-vendor cloud stacks in 2015 spent years migrating out. Don’t repeat that mistake at the agent layer.

Treat context as portable IP. Your organizational world model (process knowledge, interaction history, learned workflow patterns) is the hardest-to-rebuild and most valuable asset in the stack, so make sure no single vendor or model provider can lock it in. The right architecture separates accumulated context from the model layer so you retain your organizational IP regardless of which models or platforms you use tomorrow.

Start the flywheel early. The compounding advantage in context accrues only with deployment, so time spent evaluating adds nothing to it. Every agent execution generates organizational learning. Companies that wait to “see how it plays out” will have fewer execution traces to learn from, and they will forfeit years of compounding to first movers who build and maintain their process knowledge now.