In her article entitled AI moved forward, marketing did not, Melissa Reeve, Co-Founder, Agile Marketing Alliance, makes a valid point:
Marketers were early adopters of AI, but while it has evolved most teams are still stuck using it like a smarter autocomplete.
According to Reeve, the key reason for marketing to miss the AI race is erosion of trust with the early LLM models.
Marketers used ChatGPT to write something and it came up with hallucinations. As a result, they learned to keep genAI on a tight leash and used it for low-stakes drafts. That was rational at the time. The problem is that the lesson calcified into a habit. (If only they’d read my “what me worry” takes on hallucination instead!)
As a result, marketers failed to realize that the LLMs that they were skeptical about evolved over time. From drafting and summarizing in the fall of 2023 (GPT-4), LLMs are now capable of autonomously executing complex, multi-step tasks for hours three years later (Claude Sonnet 4.5). Most marketers have missed the evolution.
To appreciate just how far genAI has come, the author asked ChatGPT to do a fairly complex task:
I asked a current-generation model to take a published blog post, research three competitive angles I hadn’t covered, draft a follow-up piece with a different argument.
I used the same prompt on my recent blog post.
Take my published blog post https://gtm360.com/blog/2026/03/04/ai-didnt-invent-circular-deals/, research additional angles I haven’t covered, and draft a follow-up piece with an orthogonal argument.
Here’s what I got back from ChatGPT seconds later.
Circular Deals Don’t Fake Revenue – They Distort Demand
In my previous post entitled AI Didn’t Invent Circular Deals , I argued that circular deals are not new. I averred that
- They’ve been around for decades.
- They’re not necessarily round-tripping.
- And they often create real value.
All true.
But that still leaves a nagging question:
If everything is legitimate… why does it feel uncomfortable?
Because circular deals don’t just affect revenue. They affect what that revenue *means*.
Let’s see how.
1. When Investment Becomes Demand
In a normal market:
Customers create demand. Vendors fulfill it.
In a circular deal:
Investors fund a company. The company spends that money with the investor. Revenue is booked.
Nothing illegal. Nothing fake.
But something subtle has changed.
> Demand is no longer fully independent.
It’s a bit like the old company town store.
Workers are paid by the company – and spend that money at the company store.
The transactions are real. The revenue is real. But the demand is… funded.
> Capital flows out – and quietly flows back.
2. Circularity Couples the System
One circular deal is harmless. But an ecosystem of them is different.
In AI today:
The same companies are investors, suppliers, and customers. Capital and revenue move in a loop.
That loop works – until it doesn’t.
When one player slows:
- Demand falls
- Revenue falls
- Valuations fall
together.
> Circularity doesn’t just connect companies.
> It couples them.
3. This Isn’t Just Finance. It’s GTM
We treat these as financial structures.
They are.
But they’re also go-to-market strategy.
What do they actually do?
- Pre-secure a customer
- Lock in consumption
- Manufacture traction
In effect:
> You’re not just funding a company.
> You’re underwriting its revenue.That’s not a bug.
In emerging markets, it’s often the only way to get started.
Closing Thought
Circular deals are not new. They are not necessarily problematic. And they are not the same as round-tripping.
But they do something important. They blur a line we usually take for granted:
> The line between market demand and funded demand.
So the real question isn’t:
“Are these revenues real?”
They are.
The real question is:
> How much of this demand comes from the market – and how much is simply circulating within the system?
Voilà. ChatGPT delivered. First time. I have not made a single edit to the above text.
I agree with Melissa Reeve that the ceiling of what AI can do in marketing has indeed moved up. It’s down to marketers to keep up or be left behind.
As an aside, this made me wonder whether LLMs possess critical thinking ability. So I described this specific exercise and posed the question to fellow members of a PANIIT AI group.
As it often happens in gatherings of high-IQ humans, the group regularly has heated debates on several high-level intellectual themes – but, when confronted with a pointed question in a specific context, silence follows.
This time was no different. Nobody answered.
Then I asked ChatGPT the same question.
Click here to read its full answer.
tl;dr: GenAI is very good at simulating the outputs of critical thinking. And increasingly, that simulation is powerful enough to be useful in real world business.
I tend to agree with this take and look forward to hearing your thoughts in the comments below.
