SharkNinja gives one-third of staff a week to jailbreak AI

By Tom Whitfield

This week, about one-third of SharkNinja’s 4,600 employees will shut down their work computer environments for a week to experiment with AI on company problems. The global product design and technology company, known for cordless and robotic vacuum cleaners, is holding its second “Jailbreak SharkNinja” week.

The experiment directly touches a core people-process choice: whether AI work stays with technical teams or opens to finance and other functions. The first event, held earlier last spring, halted regular operations for a week, with all employees and management attempting to solve business problems through AI experiments.

SharkNinja calls the event “jailbreaking,” a term that frames participants as attacking business problems outside normal operating constraints. The label does not establish returns, however. The account gives no number of ideas, participating departments, hours saved or results from the first week.

What does the company want employees to change?

At an AWS media event in Boston on Nov. 4, CIO Velia Carboni described the purpose of the first event. She said: “Jailbreak for us was the shock of everyone being able to participate in this new kind of change that we were going through.”

She said technical groups naturally tend to move farther ahead with AI than finance, even though AI has become a normal part of technology. SharkNinja wanted to level the playing field and let people join any problem, according to Carboni.

For HR leaders, that is an adoption test rather than a skills promise. Managers must find out whether access to experiments produces better problem selection across the business, while the organization still needs evidence about which ideas work. The evidence establishes participation; it does not establish performance.

When should experiments become decisions?

AWS’s recent report, Reimagining Turning AI into Value, describes the wider issue as business redesign rather than simple workflow optimization. Tom Godden, an AWS executive in residence and one of the report’s researchers, spoke at a presentation on its findings. He said: “The constraint moved from doing to deciding, and then the deciding started out among the people.”

His observation suggests that AI changes who participates in decisions, not merely how quickly employees execute existing tasks. That matters to HR because a broad call for ideas can redistribute influence over process design. It can also expose gaps between technical and nontechnical groups, which means broad access alone does not prove readiness.

Bradley Shimmin, an analyst at Futurum Group, said many enterprises must do more than optimize workflows; they need new ways of thinking and bold changes. He called corporate inertia real, said changing course ranks among a company’s hardest tasks and noted that many organizations resist those steps.

Shimmin argued that AI can collapse the time from ideation to value realization. He also warned that organizations differ in their tolerance for failure, risk and opportunity. Some will pursue opportunities despite risk; others will not, so a copied playbook cannot settle the operating model.

Val Henderson, CEO of Caylent, a cloud services and IT consulting company and AWS partner, said the hardest part is leaving established ways of working. Her alternative is to work backward from the business as originally envisioned. Organizations should ask what they have learned, what context they have and where incremental change can produce success.

Henderson said people who rush toward an end state can make change harder to embed. For employers, that process offers a more cautious route than a blanket AI mandate: test participation, then make incremental changes where the organization can see success. The decision remains about scope, risk and whether new behavior sticks.

How can a measured adopter use AI?

The Boston Celtics, an 80-year-old professional basketball franchise, took a more measured route. It built an internal AI chatbot grounded in the AWS Bedrock AI platform, but its focus is not proving a point with AI. According to CTO and vice president Jay Wessland, the purpose is serving the goal of winning basketball games.

Wessland said the team moves in both directions on adoption. It can embrace new tools quickly and work at the forefront of analytics, but it sees no need to chase the latest technology. He said the franchise plays a human game and that technology helps it get the best players. The test is functional: does a tool improve decisions supporting that goal?

The account gives no information about chatbot users, functions, deployment scale or measured results. Nor does it say that the platform guarantees an advantage. What it does show is a company choosing usefulness over novelty, while refusing to make novelty the adoption requirement.

SharkNinja and the Celtics offer contrasting tests. One pauses work to widen participation and look for new business solutions; the other embeds a named technology where it supports a specific sporting objective. Neither example supplies the measures employers ultimately need: cost, adoption, time saved and business effect.

Until SharkNinja reports what its second event produced, the strongest conclusion is about permission, not proven value. The forward-looking question for HR is whether broad experimentation can become a repeatable way to decide, learn and improve without exposing the organization to more risk than it can manage.