Back
EP 001
27 Min
Christopher Lind has turned down multiple Chief AI Officer roles because, in his words, they are set up for catastrophic failure. After 20 years leading transformation at GE Healthcare, ChenMed, and AbbVie, he is calling 2026 the year of the AI hangover.
Most AI commentary sells potential. Christopher Lind sells the morning after. As founder and Chief AI Strategist at Christopher Lind & Co., he has spent the last year withdrawing from prestigious Chief AI Officer roles because he could see how they would end. On this episode of the Fini Podcast, he explained why so many AI strategies are doomed, what "work slop" is, and why the industry is grossly underestimating accuracy in its rush for speed.
Meet Christopher Lind
Christopher has spent over two decades leading digital transformation in environments where getting it wrong has real consequences, including GE Healthcare, ChenMed, and AbbVie. Today he advises companies as an independent AI strategist, and he is one of the few willing to talk publicly about the AI hangover, the agentic letdown, and why "made by humans" is about to become a luxury good.
The Chief AI Officer trap
Christopher walks away from Chief AI Officer roles for one main reason: unrealistic expectations. Companies treat the hire like a magician who will wave AI over their problems and make them disappear, while the actual business goals stay undefined. His tell is the job description. If the first 90 to 180 days demand outcomes no human could deliver, that is a red flag, and he is not going to take a prestigious title just to be laid off in six months. The lesson for support leaders: when your CEO launches an AI transformation, listen for whether anyone has defined what success actually is.
The agentic letdown and work slop
Christopher coined the "agentic letdown," and he is precise about the cause. The technology is capable, but tech companies are incentivized to write checks they cannot cash. A result that works 70% of the time in a controlled lab implodes in the real world, where there is no human oversight and no clean data. Worse, most companies overestimate how well they understand their own work, and AI only does what it is told. His example of "work slop": a CX team brainstormed priorities with AI, passed AI-generated decks through several teams, and got budget approved, all without ever talking to a customer. Six months of work, scrapped, because it sounded plausible enough to skate through. How do you spot it? Follow-up questions. If the person presenting cannot answer how they got there or what the risks are, send it back.
Accuracy over speed: the hangover
Christopher thinks the industry is grossly underestimating accuracy. Teams celebrated activity as if it were effectiveness, dashboards of AI adoption, faster response times, tickets closed, without checking whether quality or satisfaction actually moved. He cites research where, even when AI was slow-pitched with detailed prompts and ample information, it matched a human less than 1% of the time. A 30-second answer that is wrong damages trust more than a five-minute answer that is right, which is why winning on accuracy matters more than winning on speed. Some teams got badly burned in 2025; others, in his words, are still at the bar.
Trust is a values problem, not a tech problem
Christopher's deepest point is that AI is an amplifier. It magnifies whatever foundation you have, and many companies are not built on one, they are riffing with the tides. In the rush, teams drifted from their stated values, and that is how you get ideas like AI-driven personalized pricing, which can permanently break customer trust the moment two customers compare notes. His fix is to ground AI in real values and guardrails with human oversight at the right points, because you can fire a bad hire but you cannot fire your AI. On disclosure, he favors implicit transparency: a human-centric experience that makes clear a person is one click away, so customers never feel trapped in an AI loop.
What support leaders should take from this
Define success before you hire or deploy. If no one can say what the AI is for, no title or tool will save the project.
Respect the lab-to-reality gap. A result that works in ideal conditions will not survive messy real-world volume without oversight and clean data.
Hunt for work slop with follow-up questions. If the person cannot explain how they got the answer or its risks, it is plausible noise, not value.
Weight accuracy over speed. A fast wrong answer costs more trust than a slower right one. Measure quality, not just activity.
Anchor AI to your values. Run every flashy idea through your stated principles, and keep humans overseeing the guardrails.
Equip your people. Your team already uses AI. Find out where, and prepare them to work alongside it before you scale.
Listen to the full episode
Christopher goes deeper on the AI hangover, building customer trust, and why "made by humans" is becoming a luxury good, in the full episode of the Fini Podcast. You can follow his work on LinkedIn and at Christopher Lind & Co.
AI built for accuracy and grounded in your policies, not just speed, is what Fini is built for. Book a demo to see it in action.
Leo: Welcome everyone to the Fini podcast, where we explore the real future of AI and customer experience. I'm your host Leo from Fini. My guest today has spent over 20 years leading digital transformation at organizations where getting it wrong has serious consequences: GE Healthcare, ChenMed, AbbVie. But here's what makes this conversation different. Over the past year he's voluntarily withdrawn from countless Chief AI Officer opportunities, not because he doesn't believe in AI, but because in his words, these roles are set up for catastrophic failure. Today he's the founder and Chief AI Strategist at Christopher Lind & Co., and one of the few voices willing to talk about the AI hangover, the agentic letdown, and why made by humans is about to become a luxury good. Christopher Lind, welcome to the show.
Christopher Lind: Hey, thanks for having me Leo. Looking forward to chatting.
Leo: Let's start with a headline. You walked away from countless Chief AI Officer roles, prestigious positions with significant compensation, but you saw something that made you say, no thanks, this is doomed. Walk me through it. What are the red flags you're hearing that tell you it's going to end badly, and what should CX leaders listen for when their own CEO wants a similar AI transformation?
Christopher Lind: It's a good question, because it's not easy to walk away from a title with the word chief in front of it. But I've been around long enough to know that when things are hot, they can get cold real quick. I have eight kids, so I'm thoughtful about decisions because they affect a lot of people. The biggest one I run into is unrealistic expectations on what the role can actually do. A Chief AI Officer can be a fantastic person to have in the room, but a lot of companies think bringing in this person is like bringing in a magician who'll wave AI and magically solve all their problems. That isn't realistic, especially when you place that burden on someone who isn't responsible for most of the things involved. In the interview I'm asking the C-suite what they're trying to accomplish, and when that's not clear it's exponentially harder. So if the job description expects outcomes in the first 90 or 180 days that aren't possible for anyone, that's a huge red flag.
Leo: They just want a magician to come in and fix everything.
Christopher Lind: Right, and you already know going in you're not going to do it, so why would I do that all for a title? No.
Leo: You coined a term, the agentic letdown. We were promised autonomous agents would handle the majority of the workload overnight, and the reality feels different. Is the technology missing something, or are we just bad at implementing it?
Christopher Lind: It's a complicated mix. It's not that the technology isn't capable of doing what it should do. The problem is a lot of tech companies are incentivized to write checks they can't cash, and that's placed on people who aren't equipped to determine what's reasonable from AI. They can replicate some of their claimed results 70% of the time in ideal, lab conditions, but a real-world environment is never their best-case scenario. So a company hears it can do all these things and asks how to make it do all those things, when it was only ever designed to do a portion of them in a controlled environment with heavy human oversight. Put that in the real world and it implodes. Everyone promised potential at the start of 2025, and by the end people started waking up to what's realistic. I'd add that many organizations also overestimated how much they understood their own work. AI does what it's told, but if you don't know what to tell it because you don't have the information, it's pretty much useless.
Leo: So is it a data problem?
Christopher Lind: That's part of it, but it's also a catastrophic misunderstanding of what AI is really designed to do versus what humans are designed to do. The foundational narrative that AI can be a human is false. My friends in robotics get this, they look at the obsession with humanoid robots and see that building a bipedal robot is an engineering feat that's actually a poor way to design a robot. The biggest problem is people don't take the time to break down what AI is really good at and what it isn't.
Leo: So what does the playbook actually look like in 2026? Are you using AI to replace human agents, or to give them superpowers?
Christopher Lind: Two factors. First, there are always going to be people involved, because AI can only do what people tell it to do, it will never be truly autonomous. One of the biggest gaps today is how inconsistently equipped people are to work alongside AI. Many companies aren't preparing to equip their workforce, who are already being bombarded by 27 other AI tools. So part of the playbook is baselining and equipping everyone to use AI effectively, or you're cooked. Second, get surgical about what you're actually trying to do and how well you understand how it's done today. Once you do that, you can decide whether to replace a portion of the work or make it a superpower alongside people. I can't answer which until I get into the weeds of what you're trying to do, how you do it today, and where you're trying to go.
Leo: Are there any must-haves for AI in CX in 2026?
Christopher Lind: Not a specific tool. The must-have is a plan of attack: clear priorities. A lot of articles say you must have this or that on your punch list, but if you don't know your priorities, that's a foolish path to chase.
Leo: You talk about work slop, when AI creates noise and friction instead of helping. Give me a concrete CX example, and how much money companies are wasting on it.
Christopher Lind: Work slop will be the word of 2026. Here's a CX example. I was brought into a project where the customer experience team took customer data, brainstormed with AI on the most probable value areas, passed it through multiple teams all working alongside AI generating decks, and got all the way to approved with budget approved. And nobody had ever talked to the customers to ask if this was something they wanted. In the end they had to kill the whole thing because nothing was actually valuable, and because it had passed through so many chains of AI, nobody knew who was responsible. About six months of work, and I'm saying we might as well scrap it because none of it's real. It sounds good enough to be plausible, so if you're moving quick it skips right past you.
Leo: How do you tell genuine AI value from an expensive mess?
Christopher Lind: It may sound over-simplistic, but follow-up questions. When someone presents something and can't answer follow-up questions, send it back to the drawing board, because they clearly don't know what they're talking about and are representing something an alien intelligence came up with. People move so quick they glance at something that looks good and move forward, instead of asking how you came up with this, what would it look like, what are the risks, why might we not do this. Those questions reveal a lot.
Leo: You mentioned the AI hangover, the moment we realize speed isn't everything. Is the industry underestimating the cost of accuracy? Companies celebrate 30-second response times, but if the answer is wrong or generic, doesn't that damage trust more than making someone wait?
Christopher Lind: I wrote about the AI hangover, and I don't know that everybody realizes they're drunk yet. Some people got badly burned in 2025 because they saw activity as effectiveness, dashboards of AI adoption, reduced response time, complaints closed, but nobody looked at the quality or whether it actually improved satisfaction. Some companies partied too hard and are sobering up. Others are still at the bar. So 2026 will be an interesting year.
Leo: So the industry is underestimating accuracy over speed?
Christopher Lind: Grossly. I was involved in research comparing humans and AI, and even when you slow-pitched it with a detailed prompt and loads of information, AI did the task properly or equal to a human less than 1% of the time. We're grossly underestimating how effective this stuff actually is, and people are throwing things at it without understanding what they're throwing.
Leo: There's also a trust crisis on the customer side. If I know I'm talking to an AI that might hallucinate, my trust is already at zero. How do we build trust in an automated system, transparency or fail-safe guardrails?
Christopher Lind: Here's a good illustration. There was a time when every company had clearly stated values, and they held themselves accountable to them. In the AI rush, people forgot those existed or drifted from them on the promise that AI lets us hack it. A great example of how to destroy your customer strategy is AI-driven personalized pricing. It sounds fancy, adjust the price in real time based on what we know a customer will pay. Any CFO might love it, until customers find out you're changing prices based on them, and you'll never recover from that trust breach. If you have stated values, you run that idea through the filter and say no, we don't want two customers to discover they got radically different deals based on their zip code. AI is an amplifier, it amplifies everything you're doing, and many companies aren't built on a foundation, they're riffing with the tides, which is dangerous.
Leo: So how can support leaders build trust in an automated system and translate their values into AI?
Christopher Lind: First, make sure you actually have values and operate in alignment with them, not just words on a page. If you have them, the latest models are good at taking them into consideration as they make decisions, but that's a huge gap right now, we tell AI to make decisions without grounding its logic. There also has to be human oversight, not a person watching every action, but the right people at the right points, regularly checking whether AI is operating in alignment and staying within its guardrails, because it doesn't all the time. The difference with a bad hire is you can fire them. You can't fire your AI.
Leo: Should customers know when they're speaking to an AI?
Christopher Lind: Generally yes, depending on the situation. The key is whether you're trying to deceive. More people assume they're dealing with AI anyway. Transparency is great, but in a year of AI backlash, opening with hi, this is AI may backfire and plant a doubt that wasn't there. Better to do it in a human-centric way: I'm here to help with this, and I'll happily connect you with a live agent for anything more complicated. You didn't say I'm AI, but you communicated it clearly and reassured them they won't get trapped in an AI loop.
Leo: If you could grab a CX leader planning their 2026 AI strategy and give them one piece of advice, and one trap to avoid, what would they be?
Christopher Lind: I'll give you two. One, figure out where your current employees are with AI today, because they are using it, and if you don't know where and how, you're at major risk. Two, focus whatever you're doing with AI down to one or two things and do them really well before trying to bring AI to everything. The trap: if it sounds too good to be true, it is. Anyone who's done this role for a while knows when something doesn't pass the sniff test.
Leo: Rapid fire. The most overrated metric in CX right now?
Christopher Lind: Speed on whatever the AI is attached to.
Leo: Finish the sentence: the company that wins in 2026 isn't the one with the best AI, it's the one with the best...
Christopher Lind: People who know how to use AI effectively.
Leo: If you could un-invent one AI feature in customer service, what would it be?
Christopher Lind: The reckless chatbots.
Leo: Name one company doing AI well in CX, and why in one sentence.
Christopher Lind: One company I'm working with is automating the front line of customer interactions really well, getting people to a human the moment they need it, so AI handles a large volume of the initial stuff, customers get answers quickly, and both customers and employees are happier.
Leo: Last one. The first sign a company's AI strategy is about to fail?
Christopher Lind: They can't connect it to what they're trying to do as a business.
Leo: Christopher, this has been incredibly valuable. For everyone listening, you can find Christopher on LinkedIn or at Christopher Lind & Co. If you enjoyed this, subscribe to the Fini podcast for real conversations about AI and customer experience, not the sanitized vendor pitches. Thanks for joining us, Chris.
Christopher Lind: Thank you. Thanks for having me.
Why does Christopher Lind turn down Chief AI Officer roles?
Because many are set up to fail on unrealistic expectations. Companies treat the role like a magician who will make their problems disappear, while leaving the actual business goals undefined. When a job description demands outcomes in the first 90 to 180 days that no one could deliver, he sees it as a red flag and walks away.
What is the "agentic letdown"?
Christopher's term for the gap between what autonomous agents were promised to do and what they deliver. The technology is capable, but vendors overpromise, and results that work about 70% of the time in a controlled lab break down in the real world, where data is messy and human oversight is missing.
What is "work slop"?
AI output that sounds plausible but creates noise instead of value. His example: a CX team built AI-generated decks, passed them through several teams, and got budget approved without ever talking to a customer, then scrapped six months of work. You catch it by asking follow-up questions the presenter cannot answer.
Does accuracy matter more than speed in AI support?
Yes. Christopher argues the industry grossly underestimates accuracy. A 30-second answer that is wrong damages trust more than a five-minute answer that is right, so teams should measure quality and customer satisfaction, not just activity like response times and tickets closed.
