👋 Hey friends,

Happy Monday! I’ll be honest — I’ve been obsessing over this idea of the GenAI Divide lately. Not just the hype we see on LinkedIn about how AI is “changing everything,” but the real, messy, behind-the-scenes question: is it actually driving results?

MIT NANDA just published a massive study on enterprise AI adoption, and the headline is brutal. 95% of companies are seeing zero ROI from GenAI. Billions spent. Thousands of pilots. Endless press releases. And most businesses are still stuck in “demo land.”

In my opinion, that stat should be a wake-up call. But here’s the twist: the 5% who do get it right aren’t just experimenting — they’re crossing what the report calls the GenAI Divide, the gap between experimentation and transformation.

So today, I want to unpack this divide. Why are so many teams falling short? Why are a handful pulling ahead? And what does it really take — practically, not theoretically — to move from hype to hard ROI?

Let’s explore.

On the surface, adoption looks fantastic. I think most of us would be impressed if we saw the numbers without context:

But here’s the part that’s uncomfortable: despite all that activity, almost nothing shows up on the bottom line. Only 5% of pilots actually make it into production and deliver measurable business outcomes. That means the overwhelming majority of companies are stuck in what I’d call the “demo trap” — projects that look exciting in presentations but collapse under real-world pressure.

Why does this happen? In my opinion, it’s because most AI deployments remain surface-level. They get rolled out as add-ons or experiments, not as deeply integrated workflow changes. Drafting an email or generating a report is one thing. But using AI to reduce churn, cut agency spend, or eliminate an outsourcing contract? That’s a much harder leap, and one most organizations aren’t making yet.

Let’s zoom in on what the research actually shows. It’s pretty eye-opening:

When I look at these numbers, my takeaway is simple: the hype cycle is running ahead of the impact cycle. Businesses are eager to invest because no one wants to be left behind, but the majority don’t have the structures, workflows, or mindsets needed to turn experiments into scale. In my opinion, that gap between money spent and value created is exactly what defines the GenAI Divide.

I think the clearest way to understand the divide is by looking at two groups of companies — those stuck on the wrong side, and those crossing it.

For me, the pattern is obvious: failures chase breadth, successes go deep. The winners focus on workflows, not demos.

One of the most fascinating findings for me is the rise of what the report calls the shadow AI economy.

Here’s what’s happening: while official enterprise AI projects stall, employees aren’t waiting around. They’re buying ChatGPT Plus or Claude subscriptions on their own, and they’re using them every day to automate parts of their jobs. These tools aren’t formally sanctioned, but they’re delivering actual value — sometimes more than the official initiatives.

Think about it: nearly 90% of employees report using personal AI tools, but only 40% of organizations have rolled out official access. To me, that shows a major disconnect. The people closest to the work — the frontline employees — are finding ways to use AI effectively, while leadership is still bogged down in pilots, approvals, and RFP cycles.

In my opinion, this is a big lesson. The future of enterprise AI adoption might not come from the top down. It may come from recognizing and scaling what employees are already doing unofficially. Companies that can harness this shadow usage, rather than fight it, will move to the right side of the divide faster.

So why exactly do so many AI pilots fall apart before they scale? The research highlights one core reason, and I agree: it’s the learning gap.

Most GenAI systems today:

That makes them great for quick, low-stakes tasks — like drafting emails or summarizing documents — but terrible for mission-critical work. When the stakes are high, people want reliability, memory, and adaptability. Right now, AI just doesn’t provide that.

I think this explains why so many enterprise deployments feel frustrating. A lawyer in the study compared her firm’s $50,000 contract analysis tool to her $20 personal ChatGPT subscription — and she preferred ChatGPT. Why? Because it was flexible, responsive, and better at drafts, even though it lacked integration. That story says a lot. Until enterprise tools close the learning gap, they’ll continue to lag behind consumer tools in usefulness.

Another striking insight is where the money is flowing. Across industries, 50–70% of GenAI budgets are being spent on sales and marketing.

I’ll be honest — that doesn’t surprise me. Sales and marketing are visible functions with metrics that are easy to measure. If a tool increases demo bookings or boosts email open rates, it’s immediately clear. That makes it easier for executives to justify investment.

But in my opinion, this focus is also short-sighted. The research shows that some of the highest ROI actually comes from the back office — finance, procurement, and operations. Automating compliance checks, eliminating BPO contracts, or cutting agency spend can generate millions in savings, often with faster payback periods than front-office tools.

The problem is, those savings are harder to surface. They don’t always show up directly in revenue, and they’re less flashy to talk about. But I think the companies that lean into back-office automation will end up with stronger, more sustainable returns.

Not all startups are failing. A small group is thriving, and in my opinion, their success comes down to a few clear strategies.

First, they go narrow, not broad. Instead of trying to be everything to everyone, they pick one high-value workflow — like contract review, call summarization, or code generation — and absolutely nail it.

Second, they embed deeply into workflows. Their tools don’t sit on the side; they integrate directly with the systems people already use.

Third, they build learning systems that retain context and adapt over time. This makes them feel less like static tools and more like evolving partners.

And finally, they grow through trust networks — referrals, system integrator partnerships, and existing vendor ecosystems — not just cold outreach.

To me, this combination explains why some GenAI startups hit $1M+ ARR within a year, while others flop. The winners make themselves indispensable; the losers stay generic.

The companies that succeed on the buyer side also stand out. In my opinion, they share a different mindset: they treat AI vendors like partners, not just software providers.

Instead of asking, “What can your tool do?” they ask, “How will this tool integrate with our processes and deliver measurable outcomes?” They demand customization. They measure success on business metrics, not just model benchmarks. And they often buy instead of building, because they know internal projects fail more often than not.

What I find interesting is that successful buyers also empower the frontline. They let managers and “prosumers” — employees who are already experimenting with AI — surface use cases. Adoption starts where the need is real, then scales upward. That’s very different from centralized labs trying to push AI from the top down.

There’s been a lot of fear around AI and jobs, but the reality so far is more nuanced. According to me, here’s what’s happening:

Executives are increasingly prioritizing AI literacy. In practice, this means a recent grad who knows how to use AI tools effectively can sometimes outshine more experienced candidates who don’t. I think this is a subtle but powerful change — and it shows how AI isn’t just about tools, but about talent.

The most exciting part of the report, at least to me, is what comes next: the Agentic Web.

Instead of today’s siloed SaaS tools, we’ll move toward interoperable agents that can coordinate across the internet. Imagine agents that discover vendors, negotiate contracts, integrate APIs dynamically, and even self-optimize workflows. That’s where things are heading.

Protocols like MCP, A2A, and NANDA are already laying the groundwork. I think companies that adopt adaptive, agentic systems now will lock in major advantages — creating switching costs that competitors won’t be able to overcome easily.

Here’s how I see it. The GenAI Divide isn’t really about technology — it’s about discipline and choices. Companies on the wrong side are chasing novelty, launching pilots for the sake of appearances, and buying tools that make for good demos but don’t actually transform workflows.

The ones crossing the divide are doing the less glamorous work. They’re embedding AI into specific processes. They’re demanding tools that learn and adapt. They’re empowering the people closest to the problems, not just making decisions in central labs.

I think the real challenge in the next 18 months will be this: can leaders resist the temptation to chase hype and instead focus on workflows where AI can actually stick? Because in my opinion, the future won’t belong to the companies with the most pilots or the flashiest demos. It will belong to those who treat AI not as a novelty, but as an engine of growth — something that remembers, learns, and scales alongside the organization.

May your AI tools not just answer, but learn. May they grow alongside your business, reduce the noise, and open up the space for you to do work that truly matters.

If you could “agent-ize” just one back-office workflow in the next 90 days, which would deliver the clearest, fastest ROI for you?

I’d love to hear your thoughts — because I think this is one of the most practical and urgent questions every leader should be asking right now.

 — Naseema 

Your referral count: {{ rp_num_referrals }}

Or copy & paste your referral link to others: {{ rp_refer_url }}

That’s all for now. And, thanks for staying with us. If you have specific feedback, please let us know by leaving a comment or emailing us. We are here to serve you!

Join 130k+ AI and Data enthusiasts by subscribing to our LinkedIn page.

Become a sponsor of our next newsletter and connect with industry leaders and innovators.