Mum, can we keep the baby (AI) tiger?
Plus: The SSI pitch that worked 2 years ago doesn’t work anymore, and Agentic AI’s Biggest Consumer Moment Is Almost Here
Hi everyone, thanks for coming back to Customer Futures.
Each week I unpack the disruptive shifts around Empowerment Tech. Digital wallets, Personal AI and the future of the digital customer relationship.
If you haven’t yet signed up, why not subscribe:
Customer Futures Talks
Join the launch of MyTerms
MyTerms is a new global privacy standard by the IEEE.
Making it easier for people to express what data they want - and don’t want - to be collected online. It has the potential to replace the cookie banner, which we’d all agree is a CX fiasco.
I’ll be speaking at the launch of MyTerms on the 28th January at 8am PT / 4pm UK.
It’s a great lineup, including Doc Searls (Customer Commons) and John Bruce (Tim-Berner’s Lee’s co-founder at Inrupt).
Register for the online launch here.
The next Customer Futures Conversation: Understanding ‘OBO’
Our next conversation is a live chat on LinkedIn with Jon Evans from Digidentity. He’s the co-author of our recent OBO paper.
Where employees do things ‘On Behalf Of’ the company.
Join us on Thursday 22nd January at 4pm UK, where we’ll be talking about:
Why employee identity beyond the business is so challenging
How employee authority can leak between systems and across companies
What verifiable, time-bound authority looks like in practice
Why eIDAS 2.0, Business Wallets, and the UK DIATF make OBO inevitable
How OBO shows up in real workflows
Look forward to seeing you then.
Hi folks,
Well, 2026 has started with a bang.
More AI announcements than you can keep track of. More breaches and data disappointments than we can fathom. And yet you can still hear the quiet hum of Empowerment Tech in the background. Always building, pitching, showing up in the market in real time.
New AI agents for booking and shopping. ID wallets coming out of every nook and cranny. Data standards buried inside the new AI protocols like UCP this week from Google. And new data regulations and citizen privacy responses from governments all over the world.
But this week I’ve been mulling over something more fundamental. An age-old tension. The quiet tradeoff we experience when handing over our personal data to a product or company.
Because it’s the question we ask ourselves every day when we are presented with the ‘Please give us your email and Date of Birth’ moment at sign up.
Is it worth it?
Here’s the choice we are offered.
On one side, we can get value. We share more data, we get more convenience, better recommendations, faster services, cheaper prices or whatever.
On the other side, we can choose not do that. To build trust. To hold your data back, not provide it, to protect your privacy. But then you don’t get access to the service. Or at least not the best version of it.
Most of the time, value wins. It always has.
We give up our personal data because we want the thing. We want the free wifi, to use the app at the venue, to get the discount, or to get the 3-month free trial.
But then two things happen as a result.
First, businesses point to our always-say-yes data sharing moment and declare the case closed. “People don’t really care about their data or their privacy. If they did, they wouldn’t click ‘I accept.’”
Our data sharing behaviour becomes all the proof they need to keep gathering as much of it as possible. Consumers don’t care about privacy, they yell.
But second, people quietly start changing their behaviour anyway.
We give fake birthdays. Burner emails. Slightly wrong names and dates of birth that are fastest to enter. A phone number that ends with -0000.
Not because we’re paranoid. But because we intuitively know the trade-off is real. That once the data is taken, it will be tracked, reused, and monetised in ways we can’t see, and don’t control.
We’ve always felt that tension, even if we rarely articulated it. What feels new is that the stakes are rising.
A few weeks ago, I was offered the opportunity to “Connect your calendar to ChatGPT”.
It was one of the first genuinely “hard no” moments for me.
Not because it wouldn’t be useful - it probably would be - but because it crossed a boundary. It felt like giving root-level access to my life.
And once you notice that internal reaction, that moment in your gut, you start to see a whole bunch of questions lining up behind it.
What other tradeoffs are we making right now with our personal data? What other ‘hard no’ moments are about to kick in with businesses deploying AI absolutely everywhere? About health data? About money? About career?
The value being offered by these AI services will be there, no question. And the convenience will be compelling.
The hard question is whether the old trade-off - that we will just hand over our data - will still hold. Or whether we’re finally reaching the point where trust becomes the deciding factor?
It’s never been more important to understand what’s going on here. What the ‘value exchange’ looks like in a world of AI agents and digital wallets. In a new age of digital services, convenience and access.
And it’s never been more important to understand the future of being a digital customer.
So welcome back to the Customer Futures newsletter.
In this week’s edition:
The SSI pitch that worked 2 years ago doesn’t work anymore
Mum, can we keep the baby (AI) tiger?
Agentic AI’s Biggest Consumer Moment Is Almost Here
… and much more
Grab your hot drink of choice, a comfy corner, and Let’s Go.
The SSI pitch that worked 2 years ago doesn’t work anymore
Question of the day: if digital wallets, verifiable credentials and portable customer data are so great, and so important, why aren’t people buying? Why hasn’t the market exploded with data wallet demand?
Much of it is a comms problem. Not a value problem.
Danielle Labarbera from Iden2 has some excellent observations. An important take from the front line, and deep experience selling this stuff:
“Nobody cares about self-sovereign identity as a concept. They care that 15% of their users are abandoning signup because the KYC flow is brutal. Or that they’re hemorrhaging $2M a year to synthetic fraud that their legacy systems can’t catch.”
“The stuff that’s actually closing deals:
Selective disclosure that doesn’t add 500ms of latency. Everyone can demo zero-knowledge proofs in a controlled environment. Getting them to work in production when you’re processing millions of verifications? That’s where most implementations fall apart.
Interoperability that goes beyond marketing slides. Your credentials need to work with ISO mdoc systems, OIDC4VC flows, and yes, even the SAML dinosaurs that half the enterprise world still runs on. If you can only talk to other pure W3C implementations, your TAM is tiny.
Offline verification that genuinely works. If your credential verification fails because someone’s on a train between cities, you’ve just excluded millions of potential users.”
Then she lands the punchline:
“Privacy is nice. Reduced liability is the actual business case.
“I've watched some of the smartest people I know, real cryptography experts, lose deals because they couldn't stop talking about selective disclosure algorithms long enough to explain why it saves the CFO money.
“Meanwhile companies with honestly mediocre tech but clear ROI models are signing contracts.”
There’s a commercial truth here.
I can’t think of many (any?) businesses that are buying, or even just playing around with, verifiable credentials and decentralised identity because of ‘control’ and ‘sovereignty’.
They’re doing it because of tangible value.
Saved time. Saved money. Removed friction. More transparency and compliance. And more engagement.
You see, we’ll get all that control and privacy as a by-product anyway. But it won’t be the reason we adopt it.
The SSI pitch that worked 2 years ago doesn’t work anymore.
What story are you telling?
Mum, can we keep the baby (AI) tiger?
Gam Dias is on a roll, this week writing about AI Agents and safety.
“The tiger cub is cute right now, it’s fluffy and curious, and feels like something you could raise. A child sees a pet, but an adult sees a 500-pound predator, because that is what tigers become without exceptions.
“Generative AI is in its “cute” phase for most organisations. It’ll write the email, draft the pitch, build the landing page and cut the promo video.
“But then you connect it to real workflows. You give it memory. You let it call tools. You let it update systems. You let it transact with partners.
“And now it is not just producing content, it is shaping outcomes. It can cause real damage by accident, not because it is malicious, but because power plus complexity plus access creates new failure modes.”
Gam’s thinking is on a different level. And he’s written this practical guide to deploying agentic AI safely and scalably.
“Without wrecking customer experience or putting business continuity at risk.”
He looks at ‘the safety rails that matter when systems can act’: how you control what an agent is allowed to do, how you control what it can see and touch, and how you detect, contain, and recover when something goes wrong.
I’ve been working with Gam at our new venture, Trusted Agents, for the last few months. And we’re already working with some of the biggest brands in the world on AI agents. To dig into the powerful new customer capabilities that are about to show up on the customer side.
And I’ve never been more excited about what’s coming. More news about that soon.
Yes, our AI agents - today’s fluffy pets - are going to be much more than booking flights and handling customer services.
But that baby tiger needs to be treated like the animal it is.
Cute today, but predator tomorrow.
Agentic AI’s Biggest Consumer Moment Is Almost Here
I’m excited about a new thing.
It’s Customer Memory.
All the usual stuff I write about is quite functional. Often technical. Digital wallets and data stores and verifiable credentials. AI agents and digital IDs and portable data.
Useful, and interesting, but functional.
Customer memory is different.
It’s about creating value in our important moments. Customer experience moments. Customer disruption moments.
One of the biggest issues with CX isn’t just poor design or unnecessary friction, but not remembering me or my context when I show up. I have to repeat the same stuff over and over.
And I get treated like an imposter when I do show up, having to stand on one leg and say the magic password eight times to a secret login goblin.
Customer memory means I can bring my context with me. I can get instantly recognised and remembered.
My intent. My history. My preferences. My terms.
But sadly today it’s all product amnesia. Who are you. Why are you here. Prove it’s you.
I was particularly taken by this piece on AI Agent memory by Todd Thomas:
“What will make agentic AI feel truly inevitable for consumer experiences isn’t just better models or slicker interfaces. It’s a quieter, foundational shift called shared memory.
“Where every AI agent in a system draws from the same evolving understanding of a task, issue, or person. Think of it as a living, synchronized context layer, but one that doesn’t reset when you switch channels, doesn’t degrade when time passes, and doesn’t vanish when a different agent takes over.
“So many AI experiences feel like interns with goldfish memories. They’re capable in the moment, but they don’t reliably carry the thread forward.
“Agentic AI magnifies this issue. If agents can’t share memory, you get fragmentation at scale, with dozens or hundreds of “helpers” who don’t know what the others know.
“Once shared memory becomes standard, the most frustrating parts of digital life are going to look like artifacts from the stone age.”
Exactly right.
We didn’t know we needed customer memory until we did. We’ve been so focussed on AI capabilities, and many of us focussed on digital ID and AI Agent reputation, that we’ve missed possibly the most useful part.
Remembering what’s needed by the customer, when, why, and how to deal with it.
It’s why I’m excited about portable customer data. And what’s about to become possible when we give customers their own tools.
And most importantly, the ability for customers to be remembered.
OTHER THINGS
There are far too many interesting and important Customer Futures things to include this week.
So here are some more links to chew on:
Post: Verifiable Credentials: Why “Build Your Own Ecosystem” Is a Death Wish READ
Report: Banks must adapt the business model before third-party AI platforms capture customer relationships READ
Article: Is 2026 the year of protocols...but what are the implications for merchants? READ
News: Instagram just exposed the personal data of 17.5 million users READ
Post: Digital identity is now law in the UK — and employers need to be ready READ
And that’s a wrap. Stay tuned for more Customer Futures soon, both here and over at LinkedIn.
And if you’re not yet signed up, why not subscribe:
Spot-on analysis about the calendar moment being a watershed. What's intresting is that this isn't really a new privacy calculation, it's actually the first time the value propositon itself felt insufficent. I gave Google my email content years ago, but calendar access feels diferent because it's intention data, not just activity logs. The "customer memory" framing is clever too because it reframes portability from a compliance headache into an actual UX unlock, which is exactly the kind of shift that gets budget allocated.
Solid breakdown on the baby tiger metaphor. The part about shared memory being the foundational shift was particlarly clarifying, because most of the AI hype is still around capabilties rather than context persistence. Had a project last quarter where we tried deploying agentic workflows and the memory fragmenation became the bottleneck fast, way before any of the fancy features even mattered.