One question I get from marketing leaders is which model they should be using. Usually, I answer back by asking how knowledge moves through their team.
How the marketing leaders answer depends on their AI maturity stages. Most of the B2B marketing organizations I work with sit at the first stage without knowing it, because from the inside, stage one can look like enthusiasm.
Stage one: lots of tools
Stage one is using lots of different types of AI models and lots of different apps across your team. A copywriter pays for ChatGPT. A researcher prefers Gemini. Someone in product marketing works in Claude and will argue with you about it.
Every person picked a tool for a real reason, and each of those reasons was probably sound on the day it was made. This field moves quickly. A model that leads on creative output today is not certain to lead next month, and the incentive to go sign up somewhere else never stops.
What holds this arrangement together is a clipboard that your team ends up managing. Research comes out of one system, gets pasted into a second to be written up, then into a third to be checked. Your team has become the integration layer by moving text between browser tabs.
Your team has become the integration layer.
Stage two: one platform for everyone
Stage two is one platform for everyone. IT rolls out a single interface with several providers connected behind it, so a team can move between models without opening new accounts. Larger corporations with dedicated AI teams tend to be here already, and some have built the thing themselves and put the company name on it.
Governance improves at this stage. Spend becomes visible. Getting there is a project, though, and anyone who has run one knows the shape of it: think due diligence, data protection review, and security sign-off to name a few.
If you asked me to put numbers on it, and this is a gut estimate from the projects I see, 60% to 70% of non-technical teams are still on stage one. Most of the remainder are on stage two, and nearly all of those are large companies.
Stage three: knowledge travels
Stage three is where knowledge travels and where teams can do their most effective work. Rather than rolling out another product, you connect the expertise itself to the tools your people already have open, using a protocol that fetches it on request.
Remember the old days when documents were edited by shared file instead of shared link? When the source is shared, an improvement reaches everyone the next time they ask. When it is a copy, it starts aging the day it is made.
When it is a copy, it starts aging the day it is made.
For marketing leaders, I’d like to challenge you to ask yourself which stage of AI maturity your team is in.
What stage one costs
The first cost of stage one is uneven output. Your strongest AI users have already built sophisticated private workflows: custom assistants for specific tasks, a well-curated set of reference documents, a sequence they have refined over months. I have seen marketers do genuinely excellent work this way.
Their colleagues, doing the same job with the same subscription, produce results that need three rounds of rewriting. That gap does not close on its own.
The second cost is that the skill cannot transfer. Ask your best user to share the method and watch what happens. Inside one provider there are ways to pass a setup to a teammate. Across providers, sharing means copying. Once five people hold five copies, the next improvement will reach only one of them.
The third cost is that you cannot see any of it as the marketing leader. Work developed in private chat histories is work your organization cannot review, cannot audit, and cannot continue when the person leaves. It is also where a fair amount of unapproved tool use hides.
Ask your team, not IT
My recommendation: Before your next planning cycle, find out which stage you are actually in, and do it by asking your team rather than asking IT.
Find out who has built a workflow they rely on, and whether anyone else could run it tomorrow. Ask how many tools a single deliverable passes through. Ask where the reference material for your brand voice lives, and how someone would know it had changed.
The answers will tell you whether the expertise your team is developing stays private or gets shared.
In the end, moving along the AI maturity curve starts with knowing where you stand.