What’s up everyone, welcome to our first episode of 2025 – today we have the pleasure of sitting down with Austin Hay, Co-Founder and Co-CEO at Clarify and Martech Teacher at Reforge.

Summary: Something extraordinary is brewing in the world of martech. In the near future, Austin thinks AI agents will turn into an omniscient digital butler, anticipating your needs with uncanny precision while vanishing into the background of your workday. But the real revolution unfolds in the seemingly mundane machinery of marketing operations, where innovative companies are transforming their spaghetti mess of data pipes and platforms into something approaching digital poetry. The fundamental building blocks of our systems aren't disappearing, they're gaining superpowers. Hear it from one of our industry's most thoughtful builders.

About Austin

Austin started his career at Accenture but he left the Fortune 500 world to join a startup called Branch where he became the 4th employee
Austin then created his own boutique mobile growth engineering consultancy. He grew the practice to 1.5M with big names like Walmart, Jet, Airbnb, Foursquare and more.
His consulting practice was aqui-hired by mParticle – a leading CDP solution where he would eventually become VP of Growth
He later joined Runway as VP of Business Operations
He also started building The Marketing Technology Academy – an online learning center for martech which he would eventually sell to Reforge and become the Instructor for the new Martech course
He was also Head of Martech at Ramp, a fintech startup
Last year, Austin strapped on his jetpack and became a product founder at Clarify conquering SF and Hubspot and building the first flexible, intelligent CRM that people actually enjoy using.

AI Agents and the Hidden Promise of Ambient Computing

Let's face it, manually feeding context to AI feels a bit like teaching a fish to ride a bicycle. Current AI systems, brilliant as they may be at crunching numbers and crafting responses, still stumble around our digital workspaces like a tourist without a map. Sure, they can write a decent blog post or solve complex equations, but they're essentially working with one hand tied behind their virtual back.

Now, imagine your AI assistant as more of a digital detective, quietly observing and understanding everything happening on your screen. No more copying and pasting chunks of text or explaining what's in your Notion workspace for the hundredth time. Picture having a conversation with your computer while it maintains an almost supernatural awareness of your digital environment, from those buried Slack threads to that spreadsheet you've been avoiding. Recent demonstrations, like Kieran Flanagan's adventure with Gemini's screen reader, hint at this future, even if current versions move with all the grace of a sleepy sloth.

The real magic kicks in when we start thinking about operating system-level integration. Platform-specific AI agents are like horses wearing blinders; they can only see what's directly in front of them. But desktop applications from companies like GPT and Anthropic are pushing toward something far more interesting: AI that can understand your entire digital world, not just a tiny slice of it. It's the difference between having a personal assistant who can only help you in the kitchen versus one who can manage your entire house.

Here's where things get particularly juicy: this isn't some far-off sci-fi fantasy. We're looking at a five-year horizon where the clunky, permission-asking AI of today evolves into something far more sophisticated. The transformation won't happen overnight (sorry, instant gratification seekers), but when it does, we're talking about a 10x boost in productivity that makes current productivity hacks look like using a butter knife to cut down a forest.

Key takeaway: While today's AI assistants feel like overeager interns requiring constant supervision, the next five years will usher in truly ambient AI that seamlessly integrates with our operating systems. The future isn't about teaching AI to understand us; it's about AI that already knows what we need, when we need it, across our entire digital landscape.

The Limitations of AI Agent Marketplaces

The AI marketplace concept raises important questions about automation's role in our daily work. While downloading specialized AI agents for every task might sound appealing, reality suggests a different path forward. Current marketplace models mirror the Chrome extension ecosystem, where tools often remain peripheral rather than becoming essential to core workflows.

Austin frames the central debate in venture capital circles clearly: will we depend on AI agents that require explicit commands, or will we embrace ambient intelligence that works proactively in the background? Looking at the CRM space, Austin points out a crucial consideration that many futurists overlook. You can't simply discard two decades of sales methodology and expect professionals to embrace a completely alien interface. Instead, sellers need familiar elements: contacts, companies, opportunities, and tasks, all presented in recognizable formats that align with established workflows.

The intersection of traditional software and AI becomes particularly interesting when Austin discusses CDP platforms. Users expect certain fundamentals, like accessing persona views and tracking customer behavior. The innovation opportunity lies not in replacing these elements but in enhancing them through intelligent automation. Austin suggests that the key difference emerges in how these agents operate: will users actively assign tasks, or will agents run continuously in the background, performing expected functions without explicit direction?

While some platforms champion what Austin calls the "jack of all trades" approach with Notion-style customizable workflows, he makes a compelling case for specialized, industry-specific solutions. This focused approach, where AI agents operate autonomously within well-defined parameters, might prove more valuable than a marketplace full of generic tools. Austin emphasizes that the more specialized you are in understanding user needs, the more effective your agentic experience can be, particularly when it runs seamlessly in the background without requiring constant configuration.

The reality likely lies somewhere in between these two extremes. Certain straightforward tasks, like data enrichment for new records or basic categorization, seem well-suited for autonomous AI agents. However, more complex decisions involving customer lifecycle management, timing of promotional offers, or predictive modeling for next-best-action recommendations require deeper integration with historical data and sophisticated propensity models. The key to success may not be choosing between marketplace agents or integrated solutions, but rather understanding which approach best suits specific use cases and organizational needs.

Key takeaway: Success in AI automation won't come from marketplace-driven point solutions but through deeply integrated, industry-specific AI that enhances existing workflows while maintaining familiar interfaces. Austin's insights suggest focusing on building AI that complements rather than replaces established business processes, creating tools that feel natural rather than revolutionary.

The Core Primitives of Martech and the Path to Self-Designing APIs

Ever tried explaining Bitcoin to your grandparents? That's roughly how it feels watching companies try to skip straight to AI automation without understanding their data foundations. Austin breaks down the concept of primitives in martech, those fun...

Let's cook together: Healthy & hearty pasta e fagioli