AI, human judgement and customer experience, from a former DJ who started out on the phones.
Every Friday, my brother and I would head down to our local record shop to check out the latest tunes.
This was still the era of vinyl. CDs were the new kid on the block, and we weren’t quite ready for them yet. We’d spend hours going through the crates, pulling out 12-inch records and dropping them onto a pair of Technics 1210s. Corded headphones on. Every pop, every crackle, every hiss between the tracks.
We had a habit too. If we heard a great record built on a sample, we’d go and hunt down the original. When Biggie released Big Poppa, we went looking for the Isley Brothers’ Between the Sheets. Knowing where a sound came from made the new record mean so much more and opened up a whole new world as we’d find more tracks by those artists.
I didn’t know it then, but those Friday nights taught me a lot of what I believe about technology, people and customer experience. Including why I’m not worried about AI in the way a lot of people are.
Where I stand on AI
Let me be clear up front. I’m not anti-AI, and I’m not a techie. I’d describe myself as tech savvy.
I’ve led digital transformation, agile delivery and customer-centred design. I’ve overseen implementations of Salesforce, SAP and Avaya. I use AI myself every day, for discovery calls, interviews and structuring my thinking. I understand the power of technological change, and I’m excited by it.
But the interesting bit has never been the tech. It’s how it’s used, where it’s pointed, and who decided what good looks like.
Right now, the conversation about AI is loud. Human in the loop. Regulation. Jobs. Boards nervous about risk. Some people want to slow it all down, others want to go faster. I get all of it. But most of that noise misses the point I learned in a record shop in the 90s.
The crate and the ear
I was reminded of all this watching Rick Rubin on The Diary of a CEO. Rubin has produced everyone from Johnny Cash to Jay-Z. When Steven Bartlett asked him about AI making music, his answer went straight back to the crates.
He talked about crate digging in the early days of hip hop: going to record shops, listening to old records all the way through, and finding one short break worth building a new song on. Then he imagined AI playing music constantly in the background, and a person hearing the one clip worth using.
He also put some numbers on it. AI could make 10,000 tracks, he said, and you might like one or two of them. Maybe 5%.
That’s the whole thing for me. AI fills the crate. People find the moment.
The value was never in how many records were in the crate. It was in the ear that knew which two seconds mattered. Rubin’s view is that AI has no point of view. It’s a collection of ideas that already exist. What makes something great is someone seeing the world from a particular angle and choosing what matters.
Crate digging on the phone
At 18, I dropped out of education and went to find work. When I say find, what I mean is my mum made me.
I took a job in a contact centre, partly to keep her happy and partly because I was curious. It was a finance house providing credit for people buying white goods, cars and home improvements. The staff in the stores would call us to arrange credit for their customers.
It was never the long-term plan. But I quickly fell in love with talking to people for a living.
I spoke to a lot of the same stores, again and again, and I loved building those relationships. The small talk. Listening for small connections. Which football team they supported. Whether they had brothers or sisters. Where they grew up. Whether they were Indian too.
Those were samples. Tiny details, picked out of an ordinary conversation, that turned a routine call into a moment. And those moments built trust, call after call. I was crate digging on the phone before I knew what crate digging was.
The kit changes. The ear doesn’t.
I started DJing on vinyl. I wasn’t brilliant, but I was OK. Then along came early CDJs, with pitch bend, tempo matching and cueing built in. Then we started playing music off computers.
We went through the whole adoption cycle. Suspicion, experimenting, then wondering how we ever managed without it. The tech changed at every step. The purpose never did.
We’d still spend hours finding cue points and experimenting with transitions. The new kit made the mechanics easier. It could match the tempo and hold the cue. What it never did, not once, was choose what to play next, or feel the moment the room was ready for it. That was always the ear.
Hip hop’s whole history works like this. From Grandmaster Flash to DJ Premier and Pete Rock, every leap came from new kit in the hands of people with a point of view. The technology opened the door. The ear decided what came through it.
My career has followed the same pattern. I’ve seen Salesforce, SAP and Avaya go in. Each one was sold as the answer. None of them were. The hard part was always how people used them. A CRM can store the fact that a store manager supports the same football team as you. It can’t hear the warmth in how they say it, or know when to bring it up. The system holds the record. The ear makes it a moment.
AI is the next deck in the same cycle.
Where should organisations use AI in customer experience?
In the right place, not the loudest place. Most organisations point AI at the front line because that’s where it’s most visible, when the bigger gains are usually upstream, in the processes and systems customers never see.
Every DJ knows about the cue point: the exact spot in a record where the mix works. Drop in a bar early or a bar late and even a great record falls flat. AI is the same. Where you deploy it matters more than whether you do.
Think about the credits at the end of a film. A minute of actors, then nine minutes of people behind the curtain. Customers experience the front line, but the experience is built in those nine minutes. If they’re broken, no performance saves it. AI dropped onto the front line of a messy operation just gives you a better-lit version of the same failure.
Then there’s beatmatching. Two great records played out of time are still a trainwreck. Plenty of organisations have brilliant systems that don’t talk to each other, and teams that don’t either. AI doesn’t fix that. It puts a microscope on the cracks that people used to paper over.
And a good DJ reads the room. In most technology decisions I’ve seen, the people who actually do the job are never in the room. The tech gets built for the happy path, and done to people rather than with them.
One more lesson, and I’ve been caught out by this myself: proof of concept is not proof of viability. The demo is exciting. Then security, risk and compliance arrive late and become blockers. Bring them in on day one.
What does human in the loop really mean?
It should mean the person is the point, not the checkpoint. Too often “human in the loop” is used to describe a safety step: someone bolted on at the end to check the machine’s work so it can carry on.
In the crate, it’s the other way round. The human isn’t checking AI’s work. The human is the ear the whole thing exists to serve. So the better question isn’t where to insert a person. It’s what that person is there to notice.
There’s a harder reason judgement matters too. I once worked with an insurance client where, at month end, agents who’d hit their sales target but not their call-handling time would rush or cut calls to bring the average down. Set a target and pay to it, and people will find the loophole. But a person can also feel that something’s wrong and stop. AI has no conscience. It will chase a badly written target perfectly, including the bit you didn’t mean. That’s why the thinking has to be right before you switch it on.
Does it matter if customers know it’s AI?
Yes, because where something comes from is part of its value. Transparency rules came in this summer, but customers felt this long before any law said it.
In the same interview, Steven Bartlett told a story about a song he loved on Spotify. Two months later he found out it was made by AI, and he didn’t like it as much any more. The song hadn’t changed. What he believed about it had.
Rubin pushed back, pointing out that Steven had liked it before he knew. That’s a fair challenge, and it’s the tension every organisation now has to work through. My view is simple. Compliance sets the floor: tell people when it’s AI. Trust sets the ceiling: make the human part worth knowing about. And in my experience, transparency buys patience. Tell people it’s AI and they’ll forgive the pause.
Hip hop learned this the hard way. Samples have to be cleared. Where a sound came from matters, legally and emotionally. Just like tracking down the Isley Brothers behind Big Poppa, knowing the source changes how you feel about what you’re hearing.
Will AI take the human out of customer experience?
Not if it’s used well. The best results come from people and AI together, and the evidence is already there.
Rubin put it neatly. Fire burned down his house, but nobody would ban fire. It cooks our food and heats our homes. Powerful tools can be used well or badly. The question is never whether to use them. It’s how.
Ford is a good example. After its automated quality systems fell short, Ford brought back around 350 veteran engineers. Its VP of vehicle hardware engineering, Charles Poon, put it simply: AI is a fantastic tool, but only as good as the information you train it on. In 2026, Ford topped J.D. Power’s Initial Quality Study among mainstream brands for the first time in 16 years. And it still uses AI, now paired with the experienced people who know the work. The lesson isn’t that AI failed. It’s that AI is only as good as what, and who, you train it on.
That’s the phrase I keep coming back to: and, not or. Some of the best experiences I’ve had were purely digital. Some of the worst were purely human. It was never a versus.
Even in the debate about slowing AI down, the serious voices are arguing about pace, not stopping. Experiment hard, in safe spaces. Progress over perfection.
It was never about the decks
Looking back, I’ve been through the same cycle three times. Vinyl to CDJs to laptops. Paper to CRM to cloud. And now, AI.
Every time, the kit got better at the mechanics. It matched the tempo. It held the cue. It stored the record. And every time, the thing that made the difference stayed exactly where it always was: with the person listening for the moment.
So here’s where I land. Use AI. Use it boldly. But decide what good looks like before you switch it on, point it at the right place, and keep the people who know the work in the room. You can buy the guardrails. You can’t buy the judgement about where to point it.
AI will fill the crate with more records than any of us could ever listen to. Someone still has to find the two seconds that matter.
Because it was never about the decks. Every moment matters.
Frequently asked questions
What is the Crate and the Ear?
It’s a way of thinking about AI and people working together. AI fills the crate with options, content and data. People bring the ear: the judgement to find the moment that matters and use it well.
Is Vinay Parmar against AI in customer experience?
No. Vinay uses AI every day and has led major technology implementations. His view is that the value lies in how technology is used and implemented, not in the technology itself.
Where should AI be used in the contact centre?
In the right place, not the loudest place. The biggest gains are often in the back office and upstream processes, not just the customer-facing front line.
Should organisations tell customers when they’re talking to AI?
Yes. Transparency builds trust and buys patience, and customers care about where an experience comes from, not just how it performs.
