Not all AI experts are experts: Possibl.ai CEO

Nyssa Waters, CEO at AI consultancy Possibl.ai highlights a concerning trend she’s seen in the channel.

People are pretending to have the right AI skills to cash in on potential revenue streams, this is according to Nyssa Waters, CEO and co-founder at Possibl.ai.

Speaking to CRN Australia exclusively, she explained that while there is a skills shortage, that is only one issue.

“It's the fake knowledge rushing in to fill it,” she stated.

“There is a wave of marketing people and consultants who repositioned overnight to cash in on the craze. Confident, well-packaged, commercially motivated, and accountable for none of what they recommend. It's genuinely scary to watch.”

[READ MORE: Deep Dive: There is no AI in people, but there should be]

She highlighted two reasons why AI is becoming more a people problem than a technology problem. Firstly, she said, the platform landscape is still “incredibly fragmented”.

“The data sits in one place, the AI in another, the tools somewhere else again - and then a human has to interact with all of those messy places and stitch a workflow together across them,” she said.

“That's not a capability gap in the technology. It's a navigation burden we've handed to people.”

Second, Waters noted that working with AI is a “genuinely different mode of operating and almost nobody has been taught it”.

“The old way was linear: one task, start to finish, in sequence. AI work is multitasking at scale - running several threads at once and supervising them,” she said.

“The mental model that works best is treating it like an intern: capable, fast, and needing very clear, prescriptive direction and its work checked. Most people are still using it like a search engine, or like a colleague who already understands the context.”

However, Waters said there is an “enormous” opportunity here for partners.

“It's in a place the channel is unusually well suited to,” she explained.

Waters noted that the industry is being coy about jobs, and she said “it's costing us trust”.

“Roles are changing, some are going, and the people inside these organisations already know it,” she said.

“Reassuring messaging that dodges the question doesn't build confidence - it destroys it.

“Be specific about what changes and what's being invested in them, and you get engagement rather than quiet resistance.”

Beware the rebadge

For Waters, she believes that the rebadge is “everywhere”.

“Businesses that were marketing agencies or generalist consultancies twelve months ago now describe themselves as AI partners, with the same people, the same skills and a new tagline. Nobody retrained. The website did,” she explained.

She noted that certificates are being mistaken for capability.

“A vendor badge or a short prompt-engineering course is being presented as architecture-level expertise. Those courses teach you how to use a tool,” she said.

“They don't teach you how data should be structured, what access an agent should have, or why one model suits a task and another doesn't. That gap is where the damage happens.”

Waters blames the “demo economy” for this issue.

“It's now trivial to build something that looks impressive in a 30-minute pitch. What's hard is building something that's still working, still being used and still economic six months later. Very few of the newly minted experts have ever run a system that far,” she added.

A changing landscape

Waters explains that the world is pivoting - from a handful of large frontier models, to specialist models built for narrower tasks, to multi-model tooling where each job is routed to whichever model suits it.

She said, “In that world the model itself is not the differentiator, and it certainly shouldn't be the lock-in point.”

“What matters is the scaffolding: the agentic infrastructure underneath. How data is structured and served to agents. How agents are orchestrated and supervised. Where governance, observability and cost control sit.”

That layer is the most important element of the entire stack, and it has to be model-agnostic, according to Waters.

“It's why they built our own platform that way,” she said.

“Anyone constructing agentic infrastructure bolted to a single vendor's models is building a migration project with a delayed invoice.

“This is infrastructure and integration work. It is exactly what the channel has always been good at.”