Hosts: James & Maya
In this episode:
• Hi, I'm Maya, and today we're looking at AI showing up whether you asked for it or not — in your documents, your billing systems, and even your shoppi...
• Let's start with that first point. Google has turned Gemini
Pivot5 | 5 Headlines & Unprompted
James: Hi, I'm James.
Maya: Hi, I'm Maya, and today we're looking at AI showing up whether you asked for it or not — in your documents, your billing systems, and even your shopping cart.
James: Let's start with that first point. Google has turned Gemini on by default in Docs and Workspace. The 'help me write' prompts now appear without anyone opting in.
Maya: And the response has been frustration. There are now how-to guides circulating just to explain how to turn it off. Non-AI users have to actively disable features rather than choose to add them.
James: What stands out to me is the direction. Google is treating AI as a standard part of its core work tools, not an extra you select.
Maya: Worth noting the caveats here. Default-on is good for adoption numbers, but it tells you nothing about whether people find the features useful. It just means they're present.
James: Which brings us neatly to story two. Thomson Reuters found most professionals are now using AI at work — but far fewer are getting real value from it.
Maya: The data tells a different story than the adoption headlines suggest. The gap between use and return is widening, and the costs are concrete: lost clients, departing talent, slower professional development.
James: So the question for business leaders isn't whether to adopt. It's whether anyone is actually capturing returns from what they've already rolled out.
Maya: Exactly. Buying the tools is the easy part. The firms that close the value gap quickly will pull ahead. The ones that don't carry the cost without the benefit.
James: Story three gives us a more careful example. Mayo Clinic is putting AI across its revenue cycle — billing, coding, collections.
Maya: But here's what actually happened. Their revenue cycle chair, Todd Manion, openly said he's skeptical these functions will be fully automated anytime soon. That's a notable admission from a major health system.
James: Healthcare billing is labor-intensive and error-prone, so it's a natural fit for assistance. The interesting part is where they're keeping humans in the loop on purpose.
Maya: And other health systems will watch how Mayo balances automation with manual review before committing. Selective deployment with stated limits is a more honest model than promising end-to-end automation.
James: Now to money moving on its own. A Y Combinator startup, Agentcard, is issuing single-use virtual debit cards that AI agents can spend with.
Maya: They work anywhere Visa is accepted and connect to tools like ChatGPT and Claude Desktop. No prefunding, no wallet setup, and each transaction still requires user approval.
James: Agent-initiated buying from grocery and delivery partners is marked coming soon. They're positioning as the payment plumbing for agents — separate from what Visa and Mastercard are doing at the network level.
Maya: The user-approval step matters. It's the difference between an agent that suggests a purchase and one that completes it alone. For now, a human still signs off.
James: Which leads to story five — travel. Agentic AI can now search availability, compare prices across carriers and hotels, apply your preferences, and complete the booking with no human input.
Maya: That threatens the aggregators directly. Kayak, Expedia, and Google Flights built their business on search and conversion. An agent that bypasses those steps takes the revenue with it.
James: The open question is live inventory. Can agents reliably reach enough suppliers with current pricing to make autonomous booking genuinely competitive?
Maya: If they can, the distribution power shifts from the platforms to the agent. If they can't, the aggregators keep their position a while longer. That's the thing to track.
James: A clear thread across all five — AI is moving from optional tool to default infrastructure, and the value still has to catch up to the adoption.
Maya: Well put. The companies asking hard questions about returns, and keeping humans where judgment matters, are the ones I'd watch.
James: That's our briefing for today.
Maya: Thanks for listening, and we'll see you next time on the Pivot briefing.