Hey there,

I just finished listening to Harry Stebbings interview Andrew Ng and there were so many useful takeaways with where we are with AI tech and usage.

It made me think a lot more about how we frame AI adoption in organizations.

Most of us are getting AI adoption completely backwards.

I've been primarily talking about AI-workflows as a way to save time – and by association, costs – but this may not be the best framing.

Harvard Business Review recently published a pointed message to leadership: stop treating AI as just another efficiency tool.[1] The real opportunity, they argue, is using AI to create new revenue streams and reimagine business models entirely. EY echoes this with a stark warning: companies focused solely on cost-cutting are setting themselves up to be disrupted by competitors who see AI as a growth engine.[2]

This aligns exactly with what I've been seeing in client work. During a workshop last month, I noticed something striking: the teams most successful with AI weren't the ones focused on reducing headcount. They were the ones using AI to unlock innovation by freeing up energy for deeper thinking.

Andrew Ng highlighted this exact dynamic in the Stebbings interview, and it crystallized something I'd been observing but hadn't quite articulated. The organizations seeing true transformation aren't just replacing tasks—they're reimagining what's possible when people can focus on higher-order decisions.

Here are three approaches I've found effective when implementing AI in workflows:

IBM and Oracle's recent research on agentic AI reveals something crucial: organizations getting the most from AI are rethinking their operating models to enable "net-new" outcomes—things they couldn't do before AI, not just things they can do more cheaply.[3] Accenture's executive brief puts it bluntly: "Winners will place agentic AI where it unlocks 10x value, not 10% savings."[4]

I saw this play out perfectly with a VC investment team struggling to manage their deal pipeline. When we initially positioned AI as a summarization tool—classic cost savings framing—the response was lukewarm.

But when we experimented with Notion AI + AI Blocks to go deeper on data they already had on prospective investments, something shifted. We fleshed out the prompts to surface interesting takeaways and red flags that would be easy to miss in a manual review. Suddenly, the team was engaged.

The key was shifting from:

Same tool, completely different outcome based on how we framed the value.

Here's where many implementations go wrong. I've watched countless clients obsess over data quality and technical architecture before even thinking about adoption. But as Andrew Ng pointed out in the interview, "the biggest barrier in most large enterprises is actually people and change management. Not data."

This hit home for me because I made exactly this mistake early on. In my first Notion implementation projects, I'd spend 90% of my effort on building perfect systems and 10% on implementation.

This was completely backwards.

MIT Sloan's case studies on AI implementation back this up: the highest returns come when companies focus on transforming workflows and enabling new capabilities, not just making existing ones cheaper.[6] The technical build is increasingly commoditized—what matters is how you integrate it into the way people actually work.

I've since flipped that ratio, spending the majority of time on workflow integration and helping teams maintain flexibility while raising standards. The difference has been night and day.

One question that consistently comes up: should we implement AI horizontally across the organization or focus vertically on specific departments?

The research suggests the answer is "both, strategically." MIT Sloan's analysis shows that successful AI implementations require a nuanced approach that matches the capability to the context.[[6]]

In practice, here's what I've found works:

Horizontal approaches work well for foundational capabilities—things like document search, meeting summaries, or basic automation. These create a common baseline and help build AI literacy across the organization.

Vertical solutions drive deeper transformation in specialized functions. Each department has its own tech stack, workflows, and unique pain points. A horizontal approach often yields minimal impact because it can't address these specific needs.

The most effective implementations I've run go department by department: understand their individual challenges, map their current workflows, then craft targeted AI solutions that unlock new capabilities specific to their function.

The reality is that marketing teams need different AI capabilities than finance teams, and trying to force a one-size-fits-all approach usually means you're optimizing for the average—which means you're not truly solving anyone's problems.

The pattern across both research and my client work is clear: the most successful AI implementations focus on transforming workflows to unlock new possibilities rather than just cutting costs.

As you think about your AI strategy, consider this reframe: "What could my team accomplish if they could think deeper about decisions that were previously limited by time constraints?"

The organizations that thrive in this era won't be the ones that used AI to trim budgets. They'll be the ones that used it to unlock entirely new ways of creating value.

Would love to hear your thoughts on this topic. Have you found yourself falling into the cost-savings framing? Or have you discovered other ways to position AI as an innovation accelerator?

Until next week,
Dave