Enterprise cognition

Enterprise cognition

AI advances force every enterprise to ask a question about how to best leverage this newly available cheap intelligence for the benefit of shareholders. It is easy to get lost in the hype or completely discard the importance or magnitude of the change. I believe that even though we are entering uncharted territory, we still are better prepared than it may seem if we look to the science of cognition.

I find it useful to differentiate between intelligence and cognition. Cognition is about how systems think. Intelligence — how well they think. For understanding AI inside organizations I find it more useful to think about cognition than intelligence. Which brings me to what’s often called the 4E framework — the fourth generation of cognitive science.

My main thesis here is that cognition isn’t just something that happens inside the brain but instead it can happen in any system, as long as it is Embodied, Embedded, Enacted, and Extended. Once you start seeing cognition this way, you realise that the intelligence of a team isn’t confined to individual minds — it stretches across laptops, chats, meetings, documents, and now, AI agents.

At Togal.ai, where we build AI for the construction industry, I spend most days somewhere between code, coordination, and cognitive reflection. We use Slack, Google Workspace, AWS, Sentry, GitHub, and ChatGPT Enterprise — a pretty standard modern stack. But something interesting happens when you start to treat those tools not just as utilities, but as parts of a shared mind.

Applying AI Isn’t Like Installing a Tool

Most tech adoption follows a familiar pattern: evaluate, install, train, optimize.
AI is different. When you plug an AI model into your workflows, you’re not just adding a tool — you’re changing cognition. That means it changes not just what we can do, but how we think and coordinate.

That’s why frameworks like 4E matter. They give us language for something we feel intuitively but rarely articulate: the line between automation and intelligence is not just about speed or cost. It’s about where cognition happens in an organization.

The 4E Framework — How Systems Think

Traditional cognitive science looked for intelligence inside the brain. The 4E approach looks at how thinking happens through interaction — between bodies, tools, and environments.

Here’s how that looks in humans — and how it translates to organizations.

Once you see cognition this way, you stop asking where the “thinking” happens — and start noticing how it circulates.

Embodied: Where the Thinking Happens

At Togal, most thinking already happens inside Slack threads, GitHub comments, or Google Docs.
So when we use ChatGPT Enterprise to summarize threads or generate documentation, that’s an embodied intelligence — it lives where we work.

In other companies, this might look like Slack AI, Microsoft 365 Copilot, or Notion AI, which meet people inside the tools they already inhabit. Embodied organizational intelligence means letting the AI live among the humans — the same way cognition lives in the body.

Embedded: Feeding It Real Context

An AI without context is like a brain without sensory input.
At Togal, we’re experimenting with OpenAI’s Company Knowledge and Connectors. They let ChatGPT access company data from approved sources — Slack, Drive, GitHub — but under enterprise control. It’s still early, but it’s the closest I’ve seen to an organization building a shared working memory.

Anthropic’s Claude for Enterprise and Microsoft’s Graph Connectors are exploring similar territory — different philosophies, same goal: contextual intelligence instead of raw horsepower.

Enacted: Letting the System Learn by Doing

Cognition is not just storage; it’s action.
We’re learning that AI becomes truly useful only when it starts doing things, not just answering questions. Writing a summary, filing a bug, tagging an error in Sentry — these are all micro-actions through which the organisation “thinks.”

In this sense, AI is like a new team member who learns by participation. You don’t want it replacing people; you want it to pick up patterns of coordination by being embedded in them.

Extended: Building a Shared Mind

The more we integrate AI into our daily flow, the more the company starts to feel like a distributed mind.
ChatGPT Enterprise lets us query knowledge across different tools. GitHub stores collective reasoning about code. Slack becomes the short-term memory. Our documents and dashboards form the long-term one.

Intelligence emerges not from a single system, but from the relationships between them — the digital equivalent of neurons firing across departments.

Other companies might go further — using Notion AI for project context or Anthropic’s Claude for structured reasoning. The tools differ, but the direction is the same: cognition is being extended beyond humans, across systems.

Why This Matters

AI is forcing us to confront an uncomfortable truth:
we can’t manage intelligence as if it were infrastructure or another piece of software.
It’s not something you install — it’s something you cultivate.

At Togal, this is still an experiment. I’m trying to merge three spheres that don’t usually talk: AI, 4E cognitive science, and management. My hunch is that the next generation of companies won’t just be data-driven or AI-assisted — they’ll be cognitively designed.

When I look at our setup — the bots, the connectors, the models — I don’t see infrastructure anymore. I see a mind slowly forming between all the parts.

Maybe enterprise cognition isn’t something we build, but something we discover as we work with it. Maybe enterprise cognition isn’t something we build, but something we stumble into. And by the time we realize it, it’s already been thinking through us.H.