LogoLogo
Request a Demo

The AI Reset: When to Pause, When to Pivot, and When to Double Down

AI Knowhow: Episode

105

watch-on-youtube-final
aaple-podcast
listen-on-spotify

Episode Highlights

  • A recent study from S&P Global Market Intelligence found that nearly half of all AI POCs never make it to production
  • This finding ramps up the stakes on the question of which AI projects to pause, which to steer in a different direction, and which to shutter altogether
  • Find out the leadership mindset you need to navigate the “AI reset”

Remember when ramping up AI POCs was the hallmark of innovation? Of course you do. That was all the way back in the second half of 2024. With a new study from S&P Global Market Intelligence showing that nearly half of all AI proof of concepts never make it to production, however, it’s time to inspect how we decide which AI projects to kill, which to pivot, and which to double down on.

The 426 Project Problem

To kick off our roundtable discussion, David shares a real-world example that puts the problem into stark relief. One executive he recently spoke with claimed to be tracking 426 active AI initiatives.

“It’s a perfect example of what happens when experimentation goes unchecked,” David explains. “At some point, activity becomes the enemy of impact.”

Mohan agrees and connects it to a pattern he sees in many organizations. When teams feel pressure to prove they are keeping up with AI, they often flood the company with small, disconnected projects. While these efforts may spark short-term excitement, they rarely deliver lasting business outcomes.

Instead, David encourages leaders to identify one to three AI initiatives that are tightly linked to the company’s core strategy. These are the projects that move the needle. “Top-down initiatives are where transformation happens,” he says. “That’s where the return on investment lives.”

The message here is not to discourage experimentation but to clarify its role. Once the learning phase is complete, leaders must narrow their focus and invest where they can achieve scale and measurable results.

Navigating the Dip

This leaves leaders in a bit of a conundrum. Keep standing still and the competition will pass you by. Focus on too many AI initiatives and you’ll spread your team far too thin. So what’s a leader to do in this situation?

Every meaningful innovation goes through what Mohan calls “the dip.” Borrowing from Seth Godin’s book of the same name, the dip is the tough middle stretch of a project — too early to celebrate, but too late to quit.

“The key,” Mohan says, “is knowing whether the dip is worth the climb.”

AI projects often reach this stage when the technical challenges get hard, the data becomes messy, or the early wins fade. Leaders who are overly focused on short-term returns may lose confidence and pull back just before breakthroughs happen. Others, stuck in the sunk-cost fallacy, keep investing even when the data clearly shows the project has plateaued.

David notes that this tension is at the heart of leadership in the AI era. “The hardest part,” he says, “is stepping back far enough to make a rational decision in the middle of uncertainty.”

The best leaders create checkpoints to regularly reassess progress against business goals. They invite diverse perspectives into the conversation, looking for signs of learning and adaptation. If a project is still generating new insights, even if the ROI is not yet visible, it may be worth continuing. If it is no longer moving forward, the right choice may be to stop and redirect resources elsewhere.

So, When Do You Pivot?

The conversation ends with a few practical signals leaders can watch for as they decide whether to persevere or pivot:

  • Focus drift: Too many initiatives without clear business outcomes

  • Cultural fatigue: Teams treating AI as busywork instead of innovation

  • Data plateaus: When new inputs no longer generate useful insights

  • Momentum loss: Projects that stall in pilot mode indefinitely

Knowing which projects deserve another round of investment and which should be retired is one of the defining challenges for leaders today. The companies that succeed will not be those that do everything fast, but those that learn to move deliberately, communicate clearly, and focus relentlessly on what works.

Watch the Episode

Watch the full episode below, and be sure to subscribe to our YouTube channel.

Listen to the Episode

You can tune in to the full episode via the Spotify embed below, and you can find AI Knowhow on Apple Podcasts and anywhere else you get your podcasts.

Show Notes

You may also like