Meet the New (Customer) Bot, (Not) the Same as the Old Bot

“Who are you?/ Who, who, who, who?” —the Who (“Who Are You”)

Hey all, it’s Ian Jacobs, back on the air. I’ll be filling in for Christina McAllister for a while. For longtime readers, this will be a real throwback, as I ruled this roost for several years. In my limited run this time, I may tone down some of the snark…some. I’m still me, after all. But I also have a new gig since I last wrote here—I’m now VP and lead analyst at Opus Research, a trusted firm decoding conversational AI, voice technology, and intelligent customer interactions. So my topics will likely reflect that focus. Well, that and my random musical flights of fancy. For my first column this go ’round, I want to talk about something that’s been on my mind a lot: what happens when we as consumers start using personal AI agents, which are autonomous digital helpers managing tasks and decisions for us as individuals.

Brands, Meet Your New Customer: A Proxy With an API

Let’s just assume, for the sake of sanity, that these personal AI agents are coming. Maybe not next quarter…but soon enough that smart customer experience leaders should already be sketching out what it means to support not just humans, but their digital proxies. Unfortunately, most companies are wildly unprepared.

Sure, there will be thorny technical questions (hello, interoperability standards!), but two tech-adjacent challenges are quiet­ly lurking beneath the surface: identity and process.

First, identity. Or more precisely, authenticated delegation. If a personal AI agent shows up and says it’s acting on my behalf, how much trust does the brand extend? Does it let the agent update my address? Book a vacation? Move $10,000 out of a brokerage account? There’s no one-size-fits-all answer here. Risk tolerance will vary wildly between brands—and even between actions within the same brand.

But without some shared framework for proxy authentication, this quickly becomes a compliance nightmare. Will we see identity standards emerge from the payments world? From Open Authorization (OAuth)? Or will Apple and Google quietly replicate their iOS vs. Android turf war in the agent identity space? (Spoiler: They already are.) One way or another, companies will need a way to verify that a bot is acting with consent—and maybe even how much it knows about what it’s doing.

Second, process. Once we know who the agent is, what exactly should it be allowed to do? And how seamless should we make it? This isn’t just a permissions problem—it’s a UX and governance challenge. What if a personal AI agent is talking to a brand’s AI agent? What if it’s dealing with a human rep? Should the process be different? (Probably.) And can you even tell which kind of interaction it is in the first place?

Take a credit card dispute as an example. A personal AI agent, acting on behalf of a customer, logs into a banking app to contest a suspicious charge. It has the credentials, but the system doesn’t recognize that this isn’t the customer—it’s a bot. The dispute process kicks off automatically, and the charge is reversed. Problem is, the charge was actually valid—a recurring subscription the customer meant to keep. The AI, trained to flag anything unfamiliar, acted a little too aggressively. If the bank had mechanisms in place to recognize AI agents and limit their permissions—requiring human confirmation for certain actions, for example—it could have prevented unnecessary friction for both the customer and the merchant. Instead, everyone gets a headache.

Bottom line: If we’re building a future where customers send agents to act for them, we need to start treating those agents like real stakeholders. That means building systems that recognize, constrain, and collaborate with them—not just tolerate them.  

Ian Jacobs is vice president and lead analyst at Opus Research.

Related Articles

Automation Is a Choice, Privacy Is Under Siege, and Other Final Nuggets

Our Last Line columnist delivers his last lines for CRM.

Inclusive Experiences Start With Inclusive Language

Companies have learned and are demonstrating that words matter.

Are Robots Turning into Us, or Us into Them?

Automation technologies seem intent on turning reps and agents more and more robotic.