How AI is redefining PLG SaaS GTM with Dave Boyce @Winning by Design

saas.unbound is a podcast for and about founders who are working on scaling inspiring products that people love, brought to you by https://saas.group/, a serial acquirer of B2B SaaS companies.

In episode #37 of season 5, Tim Heicks talks with Dave Boyce, EVP Product at Winning by Design, a global B2B revenue consulting and training company that enables recurring revenue teams to architect sustainable growth.

In a recent conversation on the saas.unbound podcast (produced by SaaS Group), I dug into how product-led growth (PLG) is evolving under the influence of AI, what founders should rethink about freemium models today, and when salespeople still matter. Below I share the stories, frameworks, and practical guidance I give founders and GTM leaders—straight from my work at Winning by Design and my new book, Fremium.

From early SaaS founder to product-led evangelist

I started building software businesses back in the dot-com days and have been obsessed with the enterprise side of software ever since. Over time I watched the consumerization of enterprise software take hold—self-service, ease of use, and product-first distribution models that we now call PLG. That shift changed how companies acquire, activate, and expand their users; and today AI is amplifying that same trend.

After selling a few companies and spending time in Silicon Valley, I joined Winning by Design to help companies design repeatable, scalable GTM architectures. My focus is helping CEOs see the forest for the trees: sequencing, hiring, investment decisions, and avoiding short-term “sugar rushes” that hurt long-term scale.

What “PLG” really means (and the biggest founder misconception)

People often treat PLG like a binary: “we’re PLG” or “we’re not PLG.” That’s the wrong way to think about it. PLG is simply about making it as easy as possible for customers to buy and use your product. Having salespeople does not disqualify you from being product-led.

• Make discovery, activation, expansion, and renewal easy where you can.
• Design a self-service path where it makes sense—but also provide assisted lanes for larger or more complex deals.
• Think less about labels and more about customer empathy: does your experience minimize friction for the user who needs to get the job done?

When to add sales to a self-serve motion

We often acquire teams that built fully self-serve products with one- to three-person founding teams. Those are great starting points. The highest-leverage change is often to add a small, inbound-focused sales function that helps nurture and expand high-value leads coming from self-serve channels.

Prefer adding sales to scale repeatable, unit-based growth rather than chasing a handful of jumbo deals. I’d much rather see 50 customers paying $5k each than a single $2.5M customer—repeatability and stackable unit economics are what sustain growth.

Fremium and monetization in the age of AI

My new book, Fremium, explores premium and freemium models for today’s world. A critical reality AI brings is nonzero marginal cost: inference tokens cost real money. That changes how you think about “free forever.”

• If your product consumes paid inference, you must bound any free experience—free trials or limited usage windows are essential.
• Unit economics matter more than ever. Don’t give away the farm during a trial; design trials so users see impact, then convert.
• In some cases, charging low marginal fees from the start (metered usage, pay-per-output) makes sense.

AI accelerates empathy-driven GTM—but don’t lose the human touch

AI should be used to accelerate the parts of GTM where predictable interactions can be systematized. As Will Guidara put it, paraphrasing: “systematize the predictable so we can humanize the exceptional.” That’s the right operating principle for software teams using AI.

Here’s how to think about the robot/human/product lanes:

• Systematize predictable actions: renewals, simple configs, self-activation—use automation and bots to remove friction.
• Offer customers the choice: let them pick robot help, human help, or dive into the product on their own (Rob Gillio of Canva puts it this way: it should be the customer’s choice).
• Reserve humans for negotiation, complex stakeholder alignment, and other high-empathy situations where human judgment and relationship-building drive outsized value.

AI and discovery: the left side of the funnel is changing

Large language models (LLMs) are replacing the first-pass discovery process many buyers used to do on search engines. Instead of scanning search results, people now ask Gemini, ChatGPT, or Claude for a concise summary or recommendation.

That means content strategies must evolve:

• Think about “AI visibility” as a new kind of SEO. LLMs draw from specific sources—forums, review sites, docs—and those sources influence which products are surfaced.
• Populate the sources LLMs rely on. If an LLM leans heavily on Reddit or a particular industry site, make sure you have authoritative content there.
• There’s currently arbitrage for companies that can reverse-engineer LLM source patterns; invest early in AI-driven content strategy while tools are nascent.

Some teams are already running thousands of queries against LLMs to map which sources inform answers, then adjusting their content distribution accordingly. Expect new tools in this space—“AI visibility” trackers—that do for LLMs what SEMrush and Ahrefs do for search today.

What I look for when evaluating an acquisition target

When I help with M&A evaluation, I focus on three classic things: market, team, and product—but with sharper filters for the modern era.

• Market: Is there a real, addressable need? Size matters relative to the capital you’ll deploy—cash-efficient markets are fine.
• Team: Do the founders have curiosity, courage, and a builder’s mindset? Can they architect a scalable business rather than fake growth with one-off wins?
• Product and GTM: Is growth repeatable, stackable, and built from repeatable units (not just one big logo)? Are unit economics sound?

Avoid false positives: a big raise, a single marquee customer, or a transient growth spike aren’t reliable signals by themselves. Look for repeatable acquisition channels and sustainable unit economics.

Practical takeaways and a quick checklist

1. Don’t think in binaries—PLG is a spectrum. Make as much of your experience self-serve as possible while keeping assisted lanes for complexity.
2. If your product uses LLMs, bound free access and test monetization patterns that preserve unit economics (metering, trials, caps).
3. Use AI to automate predictable interactions; keep humans for high-empathy work where relationships and negotiation matter.
4. Invest in AI-aware content strategy: identify the sources LLMs rely on and ensure your content appears in them.
5. When evaluating acquisition targets, prioritize repeatability, stackable units of growth, and founders who can architect scalable GTM systems.

Closing thoughts

AI is not a replacement for sound GTM design; it’s an accelerant. Use it to make software easier to discover, buy, and use—while reserving human attention for the moments that truly require empathy and judgment. If you build your GTM with that principle at the core, you’ll be set for the next wave of product-led scale.

If you want to dig deeper, check out my new book Fremium and the Winning by Design materials on product-led systems. And a shout-out to the saas.unbound team for hosting the conversation that inspired this post.

“Systematize the predictable so we can humanize the exceptional.” — paraphrasing Will Guidara