Subscribe
Learn Library

Your Company's Most Valuable AI Knowledge Might Be Locked in a Document Nobody Knows About

An essay arguing that a company's most valuable AI knowledge is created by frontline non-technical staff, and that a two-way AI literacy flywheel—ignite, spread, consolidate, sustain—with owners for discovery and standardization can turn one person's prompts into company-wide standards.

ai-marketingskillworkflow
2026-08-19SupaMarketers5 min read

A while back, I sat in on a roundtable of marketers. Midway through, someone asked a friend of mine who works in content a question: how do you write creative briefs these days?

She said, I've built up a Google Doc — nine full pages of it. Brand voice, audience profiles, content formats, every kind of constraint, even the editor-in-chief's pet peeves. Every time she starts a job, she pastes all nine pages in first, then lets the AI get to work.

Does it work? Brilliantly. She uses it every single day.

Then she added, quietly: in the entire company, I'm the only one who uses this document.

My heart sank a little.

Nine pages. Carefully crafted. Used daily. Effective. And benefiting exactly one person in the whole company.

The sting got worse. A few minutes later, another participant said the same thing had happened three times at their company. Three business lines, three marketing leads, each grinding away on their own, each working out their own playbook. The result? No shared language, no shared skills, and — in his words — no "shared brain."

You'll notice that most companies don't lack people who can figure things out. What they lack is the pathway that carries what's been figured out to everyone else.

A nine-page prompt document locked on one laptop, benefiting only one person

The Heaviest AI Users May Not Be Who You Think

Guess who uses AI the hardest inside a company? Engineers? The data team?

I've seen plenty of internal usage data from clients. At one financial infrastructure company, the top ten users by token consumption were almost all in non-technical roles. Marketing, operations, customer service.

Surprising? It makes sense the moment you think about it.

Who spends all day buried in the most tedious, judgment-heavy, messiest work? They do. And the closer people sit to that kind of work, the more motivated they are to invent little tools, little workflows, little incantations — oh, right, prompts.

But this creates an awkward situation: a company's most valuable AI knowledge is being generated precisely in the widest, thinnest layer of the org chart. And the people in that layer have no channel to report what they've found upward.

Your marketing manager isn't going to file a ticket over a nine-page prompt. Why would she? In her eyes, it isn't an innovation at all. It's just how she does her job.

The One-Way Street Is Jammed

So what is the company doing about it? Most companies approach AI enablement like this: stand up a central team, pick the tools, run trainings, offer courses, publish playbooks. Knowledge flows outward from the center, reaching every employee.

Is that useful? Yes. Most people genuinely don't have time to teach themselves, so structured training is a sound investment.

But here's the problem: the central team is not where discovery happens.

If knowledge only flows out and never flows back, then every lesson learned on the front lines stays on the front lines. What one team paid tuition to learn over on the east side, another team on the west side is paying full price to learn all over again.

So AI literacy shouldn't be a "project" — it should be a flywheel that spins both ways. Ignite, spread, consolidate, sustain. One person's breakthrough should become everyone's starting point.

A two-way AI literacy flywheel: ignite, spread, consolidate, sustain

How to Get the Wheel to Spin Back

Pulling this off takes three things. Note that none of them is software.

First, you need a person whose job description says "responsible for discovery."

Call the role AI Lead, call it whatever you like. The key word is "discovery" — the duty most likely to fall through the cracks of a job description. When your marketing manager casually mentions that nine-page document in small talk, someone has to recognize it as a signal, catch it, and carry it forward. On your team right now, who is doing this? And when they hear it, where does it go?

Second, the signal needs somewhere to land.

Discovered, but buried on page three of some Slack thread — that's the same as never discovered. There has to be a named function with the authority to decide: should this discovery become standard practice across the whole company? Take that nine-page prompt, polish it, slim it down, turn it into something anyone can pick up and use. This job needs an owner, or it falls apart.

Third, experience needs a road that runs sideways.

The consumer marketing team solved a problem in March; the B2B team solved the very same problem from scratch in July. Who do you blame? You can't pin it on either team. The only thing to blame is that the company has no road that lets what one side learns travel to the other side.

What You Get Once It Spins

Winning marketing organizations have to learn faster than the outside world changes. What makes that speed possible? Precisely these channels that let one person's capability grow onto everyone overnight. When both sides of the flywheel turn, one person figures it out — and the whole company knows it.

So don't rush out to buy the next tool.

Turn back and look first. Somewhere, in some corner of some laptop at your company, a "nine-page document" is very likely sitting right now. Someone wrote it, it genuinely works, and you have no idea it exists.

Only one question remains: how do you plan to find it?

And on the day you find it, don't forget that marketing manager. It's not that she never shared — no one ever asked her.