CEOs Share Insights on AI's Impact on SaaS and Digital Commerce

Curious how successful AI SaaS founders find the real pain point they want to solve? Is it from their own workflow talking to customers seeing problems in an industry or just building and testing until something clicks? Would love to hear your thoughts

I cancelled my own webinar three weeks out. Because I realised I was teaching the wrong thing. The framework was simple: build your own AI, or buy it. Two options, clean decision. I've taught it a hundred times. But sitting with clients actually making this call, I kept watching them land somewhere outside those two options. A lot were working with AI partners and building their specific use case on top. Others were choosing to wait. Not because they weren't interested, but because their data wasn't ready and moving now meant rebuilding everything in 18 months. Build vs. Buy assumes AI is a product you either make or purchase. For most businesses it isn't. It's a set of decisions about what you own, what you outsource, what you hand to a partner, and what you're not ready for yet. So I emailed the registrants and told them I was rebuilding the session. Not a comfortable message to send. But running something I no longer believed would have been worse. The 2026 question: what do you build, buy, partner on, or pause?

You paid for the tools, you want adoption, so you publish a usage leaderboard. Your engineers then do the math. Produce enough tokens to look good and stay off your radar. Usage climbs. Value doesn't move. Ask instead: what shipped, did the quality hold, and what did each team's AI spend actually deliver. If your AI dashboard went green this quarter, do you know what it bought?

Right now, everything has an AI component. If you're running a SaaS business, investors want to know how you're using it. Vendors want to know. Even in our own conversations at G-Squared Partners, the question comes up constantly: how are you utilizing AI? We're all still in the early innings of figuring out what that actually means for each business and each industry. There's no single right answer yet, which makes it one of the more interesting things to be working through with founders right now.

Every AI GTM stack goes through the same four stages. Most teams don't know which one they're stuck on. 1) "The model will figure it out" Breaks the first time a rep sends something wrong 2) Baked into the prompt Goes stale the moment your product or the market changes 3) Static markdown files The false summit. Structured, versioned, dead within two quarters 4) Context that updates itself The only stage where AI compounds instead of drifts Julian Tempelsman wrote the field guide for GTM leaders on how to tell which stage you're in, and why the "read side" (connecting data) is the easy 10%, the "write side" (who updates it, when, and how it propagates) is the 90% nobody's solved. Worth the read if you're running GTM.

Several of the SaaS founders I've been talking to lately want to build their company AI-native from the start. They want their employees to fully embrace AI for all the work they do. Not as an optional tool sitting next to Notion. Woven into how the actual job gets done, end to end. But they are even more ambitious than that. They want those same people to build closed loops and workflows around what they do. Not only using AI as a copilot. Actively outsourcing and replacing parts of their own jobs. So they can move on to do better and smarter things. The harder work. The work only a human can do. The work most SaaS teams never get around to. This is exactly the thesis we have at Maskin, and exactly what we help companies do.

Several of the SaaS founders I've been talking to lately want to build their company AI-native from the start. They want their employees to fully embrace AI for all the work they do. Not as an optional tool sitting next to Notion. Woven into how the actual job gets done, end to end. But they are even more ambitious than that. They want those same people to build closed loops and workflows around what they do. Not only using AI as a copilot. Actively outsourcing and replacing parts of their own jobs. So they can move on to do better and smarter things. The harder work. The work only a human can do. The work most SaaS teams never get around to. This is exactly the thesis we have at Maskin, and exactly what we help companies do.

We launched our Renewal Agent last week. Today we trained two companies how to use AI to analyze their book, recommend renewal pricing strategies, and help build next year’s client budgets. It’s been interesting building in this space. A big part of it is sitting with customers, seeing how they actually use the product, understanding where things break down, and improving quickly. Another part is that there really aren’t established best practices yet—the playbook is still being written, and it’s changing fast. We’re moving quickly, gathering feedback, and iterating as we go. We’ll be sharing more as we learn and continue to build.

More users and more engagement should help your company make more money, right? While this is not exactly a trick question, the answer is more complicated than you might expect. In his Product at Heart keynote, Dr. Hayder Schneider explored some of the ways AI products behave differently from typical SaaS products, both in terms of the value they can capture and the cost it takes to keep them running. Definitely worth checking out Hayder's keynote if your company is building (or planning to build) AI products, which, judging from what I see here on LinkedIn, is approximately 100% of companies today! #productatheart

Product people tend to focus on what AI makes technically possible: faster prototyping, synthesis, and delivery of new features. But one thing that gets left out of these conversations? The true cost of AI products. In his Product at Heart keynote, Dr. Hayder Schneider explained why AI product economics isn't the same as SaaS economics. One simple example? A tweet proclaiming, "Someone on our $50 plan just ran us a $5,750 API bill." 😱 Here's how you can avoid falling into this trap: 💰 Understand that AI products have both revenue potential AND a cost-to-serve 💰 Learn which factors impact each outcome (e.g. the model you use, whether agents are involved) 💰 Don't just assume costs will go down over time—model out different scenarios. Want to get more savvy when it comes to AI product economics? Check out Hayder's keynote here: https://lnkd.in/d_yh6eMC #productatheart