Why This Nasdaq Listed CEO Changed His Mind About AI, And What It Took

Forbes

2 days ago

The climb toward AI maturity begins not with code, but with conviction.
Anyone who's survived a stint in tech or graduated from a MBA program has brushed up against the idea of maturity curves.
What emerged from the work of Richard L. Nolan and was later formalized by Watts Humphrey in the Capability Maturity Model now outlines a commonly shared intuition about how organizations grow: from chaos to structure, from instinct to optimization. But as anyone with a heartbeat in 2025 can attest, no framework quite captures the urgency that AI has dropped on the world of business.
In fact, the ecosystem resembles a pressure cooker more than neatly staged evolution.
McKinsey's latest research shows 52% of companies having established a dedicated team to lead AI adoption, and BCG has been hiring aggressively to respond to the boom in AI work. When an industry accelerates this fast, anxiety is sure to follow and a growing number of CEOs aren't asking whether to build with AI. Instead, they're wondering if it's already too late noting how the race to relevance has never been so unforgiving.
And yet, not everyone has rushed in.
Spenser Skates, CEO of Amplitude, a publicly listed digital analytics platform that serves product and marketing teams in particular, was one of the skeptics. For a long time, he wasn't sure if the AI wave was substance or just another hype cycle dressed up in neural net clothing. In October 2024, all that changed, and we can learn much from his turn from a skeptic to a convert.
Even better, we can see the results of his conversion in how Amplitude is now launching a suite of AI agents to automate product management, built directly into its platform. The first lesson is how even amidst all the hype the pivot was never inevitable.
Instead, driven by a desire to understand the value of AI for clients, the conversion process carries a lesson every executive should hear.
In a sense AI adoption has been the business world's biggest game of chicken where no one wants to be first to crash, but no one can afford to be last to move. For leaders like Spenser Skates, that tension was front and center.
'I was nervous about being too early,' Skates admits.
'The last thing I wanted was to invest heavily in something that didn't deliver value. You can burn trust that way, and it's hard to recover from launching something that hits a negative chord with clients who are looking to you for guidance,' he continues.
It's a familiar worry among founders and established CEOs alike.
Investing before the market is ready, or before the underlying technology is robust, can lead to zombie features, bloated R&D, and customer disillusionment if not the death of the entire company. This is a threat especially in the realm of SaaS, where credibility with technical users is currency and being deemed vaporware is a mortal sin.
Pat Grady, partner at Sequoia, puts it bluntly: 'Bad products have nowhere to hide. If the features aren't useful, people churn, which means whatever AI CEOs invest in, it needs to actually work for the client.'
Waiting for others to chart the path can be the safest way to ensure that what you build actually works for the client. However, it can also come at a cost and significant risks involved. In markets moving at AI's pace, the penalty for waiting too long is being outcompeted before you even get to ship your features.
Skates was acutely aware of that trade-off. 'We didn't want to be there before the value exceeded the hype, but we also couldn't afford to be late and lose the confidence of our customers,' he says.
'It was about timing something that felt inevitable but still immature. You don't want to drive a bunch of hype with customers before the product is ready to deliver an experience they're going to find valuable. We took the time to do it right before we launched Agents."
We've already seen this movie play out far beyond its trailer.
In March 2024, Cognition Labs' Devin, the self-operating software engineer, grabbed headlines as a proof-of-concept for what agentic AI could be. In the months since, a wave of agentic AI products followed: prompt-chaining copilots, automated workflows, low-code agents. Many of them, lauded at launch, now look outdated just a year later as enterprise platforms like Salesforce and emerging infrastructure like the MCP (Multi-agent Collaboration Protocol) set new bars for interoperability and performance.
Going early may be as bold as it is necessary, but going without a strategy to remain relevant when the market shifts again will get you wiped.
'We don't shy away from bold bets,' says Grady. 'But bold doesn't mean blind. The real unlock comes from timing your move with a thesis about where the value is going to concentrate, and staying nimble enough to adjust when the terrain changes.'
This insight reframes the risks of inaction for more than just Sequoia.
With AI, the bigger risk is no longer moving too soon but failing to move at all, something which Skates grappled with in his own decision to time the move.
The Amplitude leadership team, after the jump to AI
The turning point for Amplitude came when its leadership began asking a different question; not can we build something cool with AI, but would it help our customers to win?
That reframing may sound deceptively simple, but it reflects one of the most difficult moves a CEO can make: choosing to walk away from today's high-margin core in order to invest in tomorrow's unproven edge. As Clayton Christensen warned in The Innovator's Dilemma, incumbents often fail not because they miss the technology shift, but because they stay too loyal to the models, customers, and performance metrics that made them successful in the first place. Inertia is a powerful drug, but rarely a performance inducing one.
In Amplitude's case as with many other companies today, it would have been easy to keep optimizing their current platform, stay focused on power users, and leave generative AI to Sam Altman and Marc Benioff to figure out. What stands out in this turn of the maturity curve is how the future isn't going to reward AI sophistication for the sake of sophistication. Instead, it is going to reward accessibility and client-side value generation.
This mindset shift, toward long-term value over short-term certainty, is the same one we've seen behind other historical inflection points.
Apple didn't reimagine the phone because consumers were clamoring for a touchscreen device; they saw the latent potential in turning hardware into a lifestyle platform. Amazon didn't build AWS because cloud services were getting investor attention; it was a necessity born from their own scaling bottlenecks delivering unquestionable value. Paradigm shifts of the AI-kind truly begin when companies act not on consensus, but on conviction of where the value of it all is, well before the ROI spreadsheet says yes.
For Amplitude, that conviction meant going from hesitation to acceleration. And that acceleration started with an acquisition. James Evans, founder of Command AI and now Head of AI at Amplitude, puts it this way: 'When I joined, I realized Amplitude had the ingredients but not yet the recipe. Together, we built the kitchen.'
But even with the right team and the right conviction, building that recipe takes time. 'We weren't going to just slap a chat box on top of analytics and call it innovation,' Evans says. 'The hard part was embedding agents into workflows in a way that actually made people faster, not just feel futuristic.'
This is the part of innovation that rarely makes it into investor decks or product launches; the patience needed to pull it all off. Adoption, like invention, is a curve, not a click.
Which brings us to the hard part: transformation.
Most transformations fail, and according to Planview, 70% of digital transformation efforts underdeliver. Often the challenge stems from what's in the culture instead of the code.
'Real AI transformation takes more than a pilot,' says Razat Gaurav, CEO of Planview. 'It takes sustained investment, clear outcomes, and permission to fail fast. That's what separates the 30% that succeed from the rest.'
'This is particularly true for AI adoption, where the pace at which the industry is developing necessitates a more agile approach to transformation projects, including the ability to reorient them towards entirely new goals midway,' Gaurav continues.
The 70% failure rate shouldn't surprise us. If innovation were easy, we'd call it a transaction. True transformation is messy, political, iterative. It demands more than integrating a new feature or acquiring a shiny startup. Instead, it requires rethinking how your company learns, decides, and adapts.
This holds true for Amplitude as well, for which acquiring the tech wasn't the hardest part. The real transformation came from within.
'Every company that has been around long enough has technical debt,' says Francois Ajenstat, Amplitude's Chief Product Officer. 'Most also have a cultural debt. You can't bolt on AI and expect magic if you don't also update your culture. We had to rethink how teams collaborated, how we prioritized, how fast we moved. In essence, that's an entire system reboot.'
Skates agrees. 'Our purchase of Command AI and our push towards AI was ultimately an act of evolving the company.'
It is worth remembering that evolution didn't start with belief; it started with doubt.
Amplitude's story shows that skepticism, when rooted in care for the customer and clarity of purpose, is a strength instead of a weakness. Their pivot wasn't driven by hype, but by a conviction that the value was real and the timing was right. And perhaps most crucially, they treated the change not as a campaign, but as a process. A transformation, not a transaction.
For other leaders navigating the AI curve, three insights stand out. First, healthy skepticism can be a strategic asset. Skates' early hesitation gave his team the space to build deliberately, not defensively. Second, the decision to go all in came not from pressure, but from purpose, once it became clear that AI would genuinely help users win, the direction followed naturally.
And finally, Amplitude's journey reminds us that real change is infrastructural, not theatrical. It took six months of rewiring how decisions were made and how teams collaborated before the agents reached customers' hands.
As AI matures, more CEOs will find themselves in that same uncomfortable middle: too aware to ignore it, too uncertain to act.
The ones who win won't be the ones who jumped first, they'll be the ones who jumped with intent and a clear view of the value they are bringing.