Who Do You Believe When Everyone's Playing From the Same AI Playbook?

I spent a morning at a Google AI summit in Sydney, and the thing that stayed with me was not a demo. It was an echo.

Different booths. Different logos. Different brochures. One vocabulary. Agents. Data. Governance. Workflow. ROI. Dashboards. By the third pitch I could finish the rep’s sentences, and by the fifth I stopped trying.

That sameness is the story the slides keep skipping.

The pitch has converged

Google has a clean name for the centre of gravity: the Agentic Data Cloud. ^[1] The premise is sound — enterprise AI has to sit on real company data, governed properly, wired to systems that can act. Reuters reported earlier this year that Google is putting agents at the heart of its enterprise cloud push, ^[2] so this is the official direction now, not a side bet.

The convergence is real. Nearly every vendor at the event described one journey: capture the data, govern it, let the model query it, build an agent, show the result on a dashboard.

A hyperscaler, a foundation-model lab, a vertical SaaS company, and a systems integrator have genuinely different economics. Different moats, different margins, different ways to fail. They have started to sound identical.

That is the tell. The language has stopped sorting the infrastructure from the theatre.

A booth is marketing, so read it as marketing. Marketing still tells you what a vendor thinks you are ready to hear, and right now they all think you want the same four sentences. We will connect your data. We will add AI. We will make work smarter. We will show your executives the result. Each one is true. Each one could come from anyone in the room.

Every demo is a small utopia

In a demo the data is clean, the permissions behave, the customer record exists, and the workflow is obvious. The agent knows which system matters. The answer appears, the dashboard updates, the room nods.

Your company runs differently. The CRM is half maintained. The shared drive holds seven versions of one document. Sales keeps private spreadsheets. Operations has its own definition of an “active client.” A manager wants visibility and resists process change. The frontline quietly keeps the old workflow, because the new one costs them effort.

Then the AI arrives and asks one question.

“Where is your truth? When the answer is a shrug, the smartest model in the building becomes a charming intern repeating half-remembered gossip.”

The industry spent twenty years selling master data management, semantic layers, warehouses, governance tools, and transformation programs. Most companies still cannot point to one clean operating truth. The new promise is that agents finish the job. Many will stall exactly where those programs stalled: the data, the permissions, the process, the people. As AI builds fast lanes inside company red lights, the organisation still sets arrival time.

Call that the bubble risk, and be precise about the word. A bubble is a real technology pulling in too many identical promises before buyers learn which ones pay. The internet survived its bubble. So did the cloud. Enterprise AI is mid-bubble in that exact sense — the value is real, and the sameness is hiding where it lives.

What the convergence hands back

Here is the part that left me more optimistic than I expected.

The market signal is already visible: enterprise buyers are hearing the same agent story from cloud platforms, SaaS vendors, and consulting firms. Same promise. Same demo shape. Same language around workflows and automation.

When the product and the pitch commoditise — when intelligence itself meters down toward a utility — the advantage flows to the scarce layer around it. Software flattened the demo. Trust, relationships, judgement, and the willingness to stand behind a bad result kept their price.

That is consulting’s oldest ground, and abundance just made it valuable again. When every vendor can spin up the same agent, the buyer chooses on belief: who understands my business, who is still here in week six, who tells me my workflow is broken before selling me a model to hide the mess.

This is the enterprise cut of a wider shift. When everyone can build, the durable edge moves to the human layer that decides what the market notices and whom it believes. Convergence relocates that edge. It lands on the people beyond the reach of a demo. The window to spot them is open now, while the language is fresh, and it narrows every quarter.

Johnny’s verdict

Do not ask a vendor whether they have agents. Everyone has agents now. Ask what happens after the pilot — the questions that separate a partner from a pitch:

• Which workflow changes in week one?
• What has to be true in your data before the system earns its keep?
• Who owns the answer when the AI surfaces two versions of the truth?
• Does it get better from repeated use inside your business, or only in the abstract?
• What does the frontline stop doing because the system now does it first?
• What happens the week nobody opens the dashboard?

That last one is the real test. The dashboard is where AI projects go to look successful, and a system worth paying for changes behaviour before it beautifies a report.

The stack will keep converging, and describing it well is already worth nothing. The work that lasts belongs to whoever survives the mess: the people who know where the data breaks, where the workflow actually lives, who resists the change, and which small habit has to move before the big strategy becomes real.

The technology has learned how to answer. The edge belongs to the people you can hold answerable for the result.