AI Adoption for Agency Founders: Hiring, Niching, and Operationalizing

Episode SummaryIn this episode of Agency in Motion, host Tristan Pelligrino sits down with William McKee, founder of Knowmad, a HubSpot partner agency that has spent more than 20 years evolving alongside the web — and now, alongside AI. William traces the arc of his agency from a solopreneur to a focused agency serving specialty manufacturers and distributors in the chemical space. The conversation is candid about what it actually takes to keep an agency relevant across multiple platform shifts.The heart of the episode is a frank look at AI adoption inside a small agency. William shares the painful decision to let go of an entire 12-person department after a misaligned AI rollout, and how Knowmad rebuilt itself with four full-time employees doing the same revenue. He talks about the peer group he convenes monthly for agency leaders working through AI adoption, the "AI Council" and lunch-and-learn cadence inside his own company, and the realization that exposed how reliant his agency had been on knowledge trapped in individual team members' heads.Tristan and William close the conversation on a more reflective note — exploring whether this is truly the "golden age" of agencies, why technical founders may have a structural advantage right now, and the surprising limitation AI has introduced: the human judgment burden of sifting through everything it can produce. It's a peer-level conversation about reinvention, plateaus, and what it really means to operationalize an agency in motion.Guest-at-a-GlanceWilliam McKee Founder, Knowmad20+ years running Knowmad with the same business partner; HubSpot partner since approximately 2011/2012Built a niche serving specialty manufacturing and distribution clients, particularly in the chemical spaceConvenes "AI for Agencies," a monthly peer group for agency leaders navigating AI adoption (first Fridays, 2pm ET)Background as a self-described "technical marketer" dating back to building his first website in the mid-1990sFind him: LinkedIn (William L. McKee) | knowmad.comKey InsightsThe Mirror Test: When AI Adoption Fails, the Founder Is Usually the Root CauseOne of the most candid moments in the conversation comes when William describes letting go of a 12-person department after they refused to adopt an AI tool the agency had invested heavily in. His takeaway isn't that the team was wrong — it's that he was. "It's not the team, it's you. You gotta look in the mirror." For agency founders evaluating AI adoption inside their organizations, this reframes the entire conversation: misalignment on tools is almost always a symptom of how leadership introduced, contextualized, and reinforced the change. The lesson isn't to mandate harder — it's to lead more deliberately.Technical Founders Have a Structural Advantage Right Now — But It's ClosingTristan and William explore why founder-led, technically fluent agencies seem to be moving faster on AI than agencies that rely on hired specialists. The reason isn't capability — it's incentive. A hired AI strategist goes home at 5pm; a founder doesn't stop thinking about the business. That ownership mindset translates into late-night experimentation, faster pilot loops, and a willingness to absorb the learning curve personally. For founders considering whether to hire-in versus learn-in, this is a meaningful framing — especially for sub-$2M agencies where a full-time AI hire isn't economically viable.AI Exposes Where Your Agency Was Quietly FragileWhen Knowmad's head of search marketing left, the team realized just how much of their SEO process lived in one person's head. William's business partner Diana initiated a "mind meld" with the departing employee to convert tribal knowledge into an AI-driven system. The insight here goes beyond SEO: AI is forcing agencies to confront the places where they were never truly operationalized to begin with. Agencies with messy data, undocumented processes, and human-heavy workflows often find that AI makes things worse before it makes them better — because automation amplifies whatever structure (or lack of it) already exists.The Real Limitation of AI Isn't Capability — It's Judgment LoadWhen Tristan asks about AI's limitations, William's answer is striking: the constraint isn't what AI can produce, but what humans can sift. "It might've been faster to actually just write the email than to go back and forth with AI on how to craft that email." This is a quiet but important warning for agencies racing to adopt AI everywhere. The bottleneck is shifting from production to discernment — and that has real implications for how agencies design workflows, train teams, and avoid creating new forms of busywork dressed up as efficiency.