Do you remember Juicero? It was a $400 Wi-Fi-connected juicer that only worked with proprietary fruit packets, each costing as much as a juice from the store. After a much-hyped launch, the company tanked magnificently when it was discovered that the packets could simply be squeezed by hand, making the machine virtually useless.
Juicero was a classic case of a product built to solve a problem that didn’t exist. It was a shiny, expensive gadget that offered no real value. Swap “gadget” for “platform” and you have an apt description of what happens all too often with AI today. Fortunately, most companies have the internal resources to sidestep that pitfall, using what I call the “three Ds”: discuss, deputize, and dismiss.
DISCUSS
Too often, AI decisions are made behind closed doors among the higher-ups. By the time employees are brought in, the question isn’t whether AI makes sense for the organization, but which targets they need to hit with it: “Every employee must add XYZ agents to their workflow by Q3,” for example.
The result is graveyards of unmaintained agents, teams burning budget on lackluster AI applications to satisfy their bosses (tokenmaxxing, anyone?), and employees who feel AI is being done to them rather than for them. To borrow from Juicero: Ask employees whether they need the expensive appliance, or whether they can just as easily squeeze the juice by hand.
Employees will become the first line of defense against costly AI deployments when the directive to use AI comes from leadership. Managers should ask their teams where AI could make a difference in their workloads, what other tasks they could tackle if AI removed some of their to-dos, and what controls they’d need to make that work. Only when a fuller picture comes into view should the C-suite commit to major AI investments.
DEPUTIZE
The people best positioned to make AI adoption stick are those who’ve always sought out automation to free up time for focused work. Good leadership deputizes them.
At Moments Lab, our technical champion is Olivier, who runs our infrastructure. He was automating things long before agents existed: Slack routines so he’d get pinged less, self-service tools for access requests, and anything else where he spotted a repeating pattern of questions and asked himself how to make it go away. You don’t recruit that kind of person, you notice them.
Giving champions room means giving them resources and proximity to decision-making. Carve out a meaningful slice of their time, say 20%, to experiment. Include them in leadership conversations so they can shape strategy based on what they’ve learned. Ideally, pair someone with deep technical chops, like Olivier, with someone client-facing and human-centric. Between them, they’ll catch the problems worth solving and the solutions people will use.
Finally, make their work visible. Open an internal channel where champions share what worked, what failed, and what surprised them. Colleagues who see a tedious process eliminated or an hour reclaimed will want in. Leadership sets the direction; champions make it contagious.
DISMISS
Named by economist Richard Thaler and first studied by psychologists Hal Arkes and Catherine Blumer , the sunk-cost fallacy is the tendency to keep investing in something that’s failing because of how much you’ve already put into it.
With AI, it tends to take the form of zombie deployments: the pilot that’s been “three months from proving itself” for a year and a half, the enterprise license renewed because canceling would mean admitting a mistake, or the fleet of agents no one maintains. The spend becomes its own justification.
Other failures are harder to spot because they look like success. We built an autonomous agent to post a Monday recap in Slack. Its first source of truth was our company values, and since our first value is optimism, it started reframing lost deals as good news, until the sales team objected. The same agent later wrote a thank-you message to a departing colleague, drawing on their work across Slack and Notion, before any human had said a word. The message was warm and accurate, but letting a machine say goodbye first sent the wrong signal from leadership.
Elsewhere, agents raised our pull request volume so much that our most senior engineers became full-time reviewers and stopped building. And an AI-generated summary that is sent without being read by a human first simply pushes the work onto the recipient. At scale, the daily volume of text to reread becomes a management problem in its own right.
The way out is to decide what failure looks like before you deploy anything. Set clear success metrics up front, put a review cadence on the calendar, and give someone explicit authority to pull the plug. Killing an underperforming tool isn’t a defeat; it’s clearing the counter for something that works. Nobody’s obligated to keep buying Juicero packets just because they paid $400 for the machine.
AI PROJECTS GATHERING DUST
Gartner estimated earlier this year that around 50% of generative AI projects end in failure, shelved next to the Juicero and countless other solutions started in earnest but without proper planning. Sticking to the three Ds lowers the odds of choosing the wrong path and then throwing good money after bad. So grab your team a round of (real) smoothies, have those conversations first, and then start building.
Phil Petitpont is cofounder and CEO of Moments Lab.
