How GoTu Cut Marketing Spend 78% and Still Grew

In this episode of The Marketing Front Lines, we speak with Thalía Diedrick, VP of Marketing at GoTu — a dental industry marketplace connecting practices with dental professionals. Thalía took her marketing team from 13 people to 3 and simultaneously cut quarterly spend from $1.2M to $260K, a 78% reduction. Rather than treating that contraction as a crisis, she used it as a forcing function to rebuild the entire marketing function around AI-native infrastructure — and then helped close a Series A in the process. This conversation covers the hard internal conversations that actually made it work, the "modular infrastructure" framework her team uses to manage AI tools without chaos, and what it looks like to build the marketing org of the next five years right now.Topics Discussed:How a 78% budget cut became the catalyst for building an AI-native marketing orgEliminating stigma around AI use within a small team: the internal culture shift that preceded everything elseMoving from headcount reduction to capability expansion: what "AI-native" actually means operationallyBuilding a "modular infrastructure" for AI tools that survives model churnDaisy-chaining specialized AI functions — including a compliance skill, email writing skill, and AI Scrum agentThe ROI framework GoTu uses to evaluate every AI build: time savings vs. revenue generationWhy people hold AI to a higher standard than human processes — and why that's irrationalHow career growth changes for marketers when team size stops being the proxy for seniorityBuilding with Open Claw agents, Claude Code, and Lovable as a non-technical marketerLessons for B2B Marketers:Destigmatize AI use before you deploy any tools. Two and a half years ago, when Thalía took over as interim head of marketing, her first move wasn't a tool rollout — it was a cultural reset. She told her team of three to do everything with ChatGPT and removed the shame around it. People were embarrassed that AI was doing "their job." Thalía's framing: at this team size, not using it just slows everyone down. That permission structure preceded any meaningful adoption. If your team is still hiding their AI usage from each other, the bottleneck isn't the tools.Architect for model churn, not for today's favorite LLM. GoTu calls their approach "modular infrastructure" — and the explicit design goal is plug-and-play portability. They've already cycled through custom GPTs, Google Gems, and Claude skills. Their solution: document the use case, the context, the brand rules, and the expected output, not the specific model. The LLM is the function; the documented instructions are the institutional knowledge. When the next model supersedes the current one, the swap is a five-minute configuration change, not a rebuild.Daisy-chain specialized functions rather than building one monolithic AI tool. GoTu's content workflow runs through sequential, purpose-built skills: a compliance skill flags legal concerns first, then the output flows into a blog writing function built on their brand voice and reference materials. Thalía's principle: the more specialized the function, the better the output. Trying to build one AI that does everything well produces mediocre results across the board. Build narrow, document thoroughly, and chain them together.//Sponsors: Front Lines — We help B2B tech companies launch, manage, and grow podcasts that drive demand, awareness, and thought leadership. www.FrontLines.ioThe Global Talent Co. — We help tech startups find, vet, hire, pay, and retain amazing marketing talent that costs 50-70% less than the US & Europe. www.GlobalTalent.coDon't Miss: New Podcast Series — How I Hire. Senior GTM leaders share the tactical hiring frameworks they use to build winning revenue teams. Hosted by Andy Mowat, who scaled 4 unicorns from $10M to $100M+ ARR and launched Whispered to help executives find their next role. Subscribe here: https://open.spotify.com/show/53yCHlPfLSMFimtv0riPyM