Challenge

Licenses alone don't unlock AI — context does

Rolling out AI meant more than handing out seats. The real unlock was context — and our best context was sitting unused in forgotten docs and decks.

When we started rolling out AI at AXS, I didn't want to just give licenses out and expect productivity gains. I wanted my team to fully unlock the power of the tools.

But my team had varying levels of experience with AI. And I knew that that power was unlocked by giving AI context, not by spending time teaching prompt engineering.

So, what context did every team have at their disposal? Docs and decks.

You know the ones — the decks we'd pour time into making, presenting, and circulating, just to have it forgotten a week later and sit in the cloud collecting digital dust. New team members got hit with a wall of information on day one, most of which they'd forget by week two.

When someone did remember a doc existed, finding it meant tracking down whoever had the link, and waiting. Hours lost, sometimes days. Then came the real work: reading it, finding the part that applied to them, and interpreting it.

Solution

Turn process docs into GPTs — a 24/7 coach for the team

Diagram of AI agent capabilities for writing, thinking, researching, collaborating, and synthesizing

Prompt engineering fades. Context doesn't. I turned our docs into purpose-built GPTs so everyone could work from the same level of quality.

Early on, I read a lot about "prompt engineering" — the art of structuring the perfect prompt. I knew we'd look back and laugh at that in a year or so. AI was getting smart enough to make it unnecessary, and it's already fading. What doesn't fade is context.

It's why I can say less to my best friend and still be understood — we share history. AI works the same way. And the way you provide that history from day one is by feeding it context.

I started experimenting with Projects and GPTs in my own hobbies, and it felt like having a 24/7 coach — that same shorthand you get with an old friend.

I wanted everyone on my team working from the same level of quality. So I turned our docs into GPTs: a UX copywriting expert trained on our tone and voice, a research assistant that covered planning through synthesis, and a Career Coach built on our Career Ladder. Mini chat agents that could ensure a certain level of quality in our work.

The goal was never to replace the work. It was to raise the floor on everyone's output. That's what makes the way I like to roll out AI empowering instead of threatening.
Results

Hours saved, new skills, and company-wide adoption

Survey results charts showing strategic thinking, brainstorming, time savings, and quality improvements
9purpose-built agents built by my team
$200k+Annual retainer saved on UX copywriters
100%Design team adoption

Bi-weekly surveys showed hours — sometimes days — saved. The team pitched their own GPT ideas, and after a workshop, other teams across the company followed.

As the first team at AXS to do this, I tracked usage with a bi-weekly survey. People reported saving hours, sometimes days, and said they felt like they'd picked up new skills. They started pitching their own GPT ideas, which we built and shared with the team.

After I ran a workshop on AI and agents, other teams across the company started building and sharing their own.