People & Prompts: Why AI Pilots Are Failing, and Sales Still Needs Humans

By Darrell Bowman, Search Partner

Part of our “People & Prompts Series” about the impact of AI in the workplace and the future of work.

A recent study from MIT found that 95 percent of generative AI pilot projects in companies are failing to deliver meaningful results. Despite all the hype, most organizations are struggling to generate ROI from their AI investments. The problem isn’t the technology itself. It’s where and how companies are choosing to deploy it.

Companies are hesitant to integrate AI into mission-critical systems like accounting, logistics, or compliance. These areas are deeply embedded, high-risk, and expensive to change. Many have already undergone complex transformations to reach their current state. The memory of those difficult transitions, including layoffs, budget overruns, and operational disruptions, makes teams understandably cautious. When something works, no one wants to be the person who breaks it.

Instead, organizations are piloting AI in low-risk, low-impact areas, especially in sales and marketing. They test it in places where mistakes won’t affect existing customer relationships or operations. This usually means top-of-funnel activities like outbound emails, prospecting bots, or basic customer service interactions. These are safe zones. But ironically, they are also the places where people matter most.

AI might be able to write an email or answer a question, but the top of the funnel is where trust is built. First impressions still matter, and clumsy automation does more harm than good. It’s one thing to get ignored by a prospect. It’s another to be dismissed because you sounded like a robot.

Here’s the deeper issue: AI is designed to give the safest, most statistically probable answer. It works by identifying patterns across large data sets and selecting the response that best reflects the center of the distribution. But new ideas, especially the ones that are truly innovative or high-impact, don’t come from the center. They live in the long tail, in the edge cases and outliers. This is where humans thrive.

Great salespeople don’t just repeat what’s worked before. They challenge assumptions, read between the lines, and pick up on the subtle cues that signal an opportunity. They hear what’s not being said. They pivot in the moment. And they do it all in real time. These are not tasks that can be easily scripted or automated. They require emotional intelligence, strategic thinking, and situational awareness. These are skills that AI simply doesn’t possess.

Marketing is no different. The best campaigns aren’t generated by prompts. They come from deep human insight: understanding the buyer’s psychology, identifying hidden motivations, and knowing when to break the rules. AI can be a useful assistant, but it cannot originate strategy. It can remix ideas, but it cannot generate new ones with context and meaning.

Now that the economy is showing signs of contraction, some leaders are eyeing AI as a way to cut costs. Scaling back on people and betting on automation instead may sound efficient, but it’s a dangerous bet. Cutting your top-of-funnel team and replacing them with a prompt stack is like firing your best players and handing the game plan to an intern with a spreadsheet.

AI is a tool. It can help with efficiency, summarization, and scale. But it cannot replace the strategic advantage of a strong human team, especially in sales, marketing, and customer engagement. If 95 percent of pilots are failing, that’s not a technology problem. That’s a deployment problem. It is often a leadership problem too.

As you plan your next move, ask yourself: Are you using AI to support your people, or to replace them? Are you trying to eliminate complexity, or understand it better? The companies that win in this next cycle won’t be the ones who automate first. They will be the ones who know when not to.

Because when it comes to building trust, spotting opportunity, and creating real customer value, people still outperform prompts. Every time.

 Feel free to share your thoughts with me via email

Previous
Previous

Prompts & People: Leading Engineering Teams Through Change

Next
Next

Risk Capital Isn’t What It Used to Be: The Harsh Reality for Canadian Tech Startups in 2025