Most Work Is Not Content

Ask anyone what they use AI for and the answer is almost always the same. Writing emails. Drafting reports. Generating marketing copy. Summarizing documents. Writing code. The entire commercial AI ecosystem orbits around one idea: produce text faster.

AI is genuinely good at producing text. But somewhere along the way, we confused "good at generating content" with "good at doing work." Those are not the same thing.

Look at what people actually do in their jobs. Yes, some of it is writing. Emails, presentations, spreadsheets, documentation. But a massive portion of actual work has nothing to do with producing content. It is coordination. It is decision-making. It is reviewing, approving, escalating, triaging. It is noticing that something is wrong before anyone reports it. It is maintaining relationships, managing expectations, navigating ambiguity. It is the thousand small judgments that never become a document.

We have built an entire industry around the fraction of work that happens to be text-shaped. And then we wonder why AI adoption stalls once you move past the early wins.

The problem is the mental model. We still think about AI through the lens of traditional automation: find a repetitive task, automate it, measure the time saved. That worked for factory floors and data entry. It even works for content generation, because writing an email is a discrete, repeatable, output-oriented task. But it falls apart the moment you try to apply it to work that is not about producing an artifact.

Consider a project manager. Their value is not in the status reports they write. It is in the fact that they know which conversation to have with which person at which moment to keep a project from derailing. An account manager's value is not the emails they send. It is the relationship context they carry in their head and the judgment calls they make about when to push and when to wait. A senior engineer's value is not the code they write. It is the code they prevent from being written.

None of that looks like content generation. And because it does not, we have no idea how to point AI at it.

We keep asking "what content can AI produce?" when we should be asking "what decisions can AI inform, what patterns can it notice, what coordination can it handle?" The highest-value work in most organizations is not about output volume. It is about signal detection, prioritization, and timing. An AI that can draft a perfect email is useful. An AI that knows which email not to send is transformative.

The companies that figure out how to apply AI beyond content generation will have an enormous advantage, because the framing is different. They will be solving problems that everyone else has not even named yet.

We are stuck in the content era of AI. The agents era should be about something bigger. Not producing more, but understanding more, coordinating more, and deciding better. The unlock is better perception.