Claim OwnershipKnowledge Distillation Podcast
Subscribed: 1Played: 2
Subscribe
© Copyright by ASK-Y
Description
Knowledge Distillation: The Rise of the AI Analyst
Welcome to Knowledge Distillation – a series exploring how AI Analysts are transforming the future of data work. We look at practitioners becoming AI Analysts, founders building AI Analyst tools, VCs backing the AI Analyst wave, and market analysts mapping the trend.
Each episode uncovers what it means to be an AI Analyst today – the workflows being reinvented, the skills analysts need now, and the promises AI is keeping or breaking. From prompt engineering to context management, we dive into the real conversations shaping this role.
Let’s distill some knowledge. Because bots won’t win. AI Analysts will.
Explore More: www.ask-y.ai
15 Episodes
Reverse
John Lovett has seen every chapter of digital analytics from the inside. He started as a marketer curious about how messages reach people, moved through analyst roles at Jupiter Research and Forrester, co-founded Web Analytics Demystified with Eric Peterson, served as president of the Digital Analytics Association during the pivotal renaming from “web” to “digital” analytics, and is now VP of Analytics & Insights at Seer Interactive, where he’s building AI into every layer of the analytics practice.
In this episode, John and I dig into what actually makes an analyst an AI analyst – and his answer surprised me. It’s less about new skills and more about bringing AI into the curiosity and critical thinking that always made someone a good analyst in the first place. He walks us through how Seer identified 15 core deliverables and systematically disrupted each one with AI, creating a team that starts every task with an agent rather than a blank template.
We then dive deep into agentic commerce – the emerging reality of AI agents that browse, compare, and buy on behalf of consumers. John shares what Seer is learning about distinguishing human from bot traffic (spoiler: humans do “clicky clicky scrolly scrolly,” agents do surgical strikes), why log file analysis is making a comeback, and his hypothesis on which product categories will see agentic purchasing first.
Finally, John walks us through his groundbreaking GEO research using the 2026 Winter Olympics as a case study – running 5,000 prompts across every major LLM to understand how models find, trust, and cite brands. The results reveal a fascinating divide between models that search the web and those that hit what he calls “the binary cliff.” Plus, we talk about his new book, The New Big Book of KPIs, and why the best time to start with AI would have been yesterday – but the next best time is today.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant speaks with Simo Ahava – quite simply the person the entire digital analytics and technical marketing community turns to when they need to understand how things actually work. Simo has been writing about web analytics, tag management, and the Google marketing stack since 2010, and his blog at simoahava.com has become the definitive technical reference for anyone implementing Google Analytics or Google Tag Manager. A Google Developer Expert in both platforms from 2014 to 2025, a multiple Digital Analytics Association award finalist, and one of the most generous knowledge sharers the industry has ever seen – if you’ve ever asked a question on Measure Slack, there’s a good chance Simo answered it, thoughtfully, for free. He co-founded Simmer with his wife Mari Ahava, an online learning platform for technical marketers that has become the gold standard for courses on server-side tagging and BigQuery. He is partner and co-founder at 8-bit-sheep, a Helsinki-based digital services consultancy, and co-hosts the Standard Deviation Podcast with Juliana Jackson.
The conversation opens with what Simo calls the educator’s dilemma: AI makes it trivially easy to get answers, which removes the incentive for deep learning. His students take course content to an LLM, get a conflicting answer, and bring the contradiction back – without the baseline knowledge to judge which is correct. Katrin pushes back: practitioners doing real analytics work need to understand fundamentals like context windows and attention mechanisms. They land on a distinction – Simo’s concern applies to learners seeking quick answers, Katrin’s to practitioners maintaining context continuity across complex workflows.
The episode then pivots to agentic commerce. Simo draws a direct line from his data layer and server-side tracking expertise to the challenge of designing websites for AI agent access. Tag management systems have let organizations survive with poorly structured data for years. Agentic commerce breaks that – agents need structured data by design, not retroactive patches. Simo warns against over-optimizing for agents at the expense of human UX, and raises the unsolved measurement problem: how do you track agentic traffic when AI agents have no reason to identify themselves?
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant speaks with Scott Brinker – the creator of the Marketing Technology Landscape Supergraphic, the map of the martech industry that started with 150 logos in 2011 and now tracks over 14,000. Scott spent eight years as VP of Platform Ecosystem at HubSpot, where he built out their partnership and integration ecosystem. He holds degrees from Columbia and MIT Sloan, co-authors the annual State of Martech report with Franz Riemersma, and now works full-time as an independent martech analyst through Chief MarTech. Katrin still drinks her coffee from a mug Scott gave out a decade ago – the one with snails at a boardroom table and the tagline: technology changes exponentially, organizations change logarithmically.
Together they dig into the so-called SaaS Apocalypse – triggered by AI-native tools lowering the barrier to building software – and land on a nuanced take: the market overreacted in the short term, but the long-term disruption to SaaS business models is real. The risk isn’t that customers will vibe code their own CRM; it’s that a thousand new companies will. Scott introduces his framework of systems of truth and systems of context – an evolution of the classic systems of record and systems of engagement – and explains why delivering the right information, to the right person or agent, at the right moment is the hardest and most valuable problem in martech today. Katrin connects this directly to Ask-Y’s thesis: that the central challenge in analytics isn’t the tools, it’s maintaining context continuity across every step of the workflow – from data connection through transformation to stakeholder output.
The conversation goes deep on Scott’s framework of three types of AI agents in marketing: agents for marketers (internal productivity), agents for customers (brand-controlled interactions like AI-powered chatbots and SDRs), and agents of customers (the disruptive category – AI assistants that work for the buyer, not the seller). They explore how agents of customers are forcing a rethink of everything from SEO to email marketing to e-commerce, and Katrin lays out her thesis that agentic commerce will trigger a workstream comparable to mobile platforming, the GA4 migration, and a fundamental shift in customer relationships – all at once. Scott agrees and adds his prediction that agentic email is the next major disruption most marketers aren’t preparing for. The episode closes on the AI analyst role itself: Scott argues that hands-on experience with AI tools is non-negotiable, that understanding code remains critical even when you’re not writing it, and that the only way to build the mental model required for this era is through consistent, daily practice. His advice: the only way out is through.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant speaks with Eliot Durbin, General Partner at Boldstart Ventures – one of enterprise software’s most active inception-stage funds, with a portfolio that includes Clay, Snyk, Wiz, Crew AI, Kustomer, and Keycard, among others. Boldstart was founded in April 2010 with a $1M first fund and pioneered what Eliot calls “inception investing”: backing technical founders on the strength of a person and a thesis – before a product, before a pitch deck, sometimes before there’s even a market. Katrin and Eliot have known each other for 15 years, with Boldstart backing Ask-Y at its earliest stage.
Together they unpack the so-called SaaS Apocalypse – the trillion-dollar collapse in software market cap triggered by AI-native competition – and whether the hype matches the reality. Eliot argues it doesn’t: software isn’t dying, it’s evolving, just as it did through the cloud and mobile revolutions. The companies that move fast and go AI-native will survive; those that don’t will go the way of the ones that missed mobile. The conversation goes deep on what actually compounds at inception in a world where anyone can vibe-code a prototype in a week, how moats are being redefined around trust and interaction data, and why speed remains the only real advantage at the earliest stage. They also dig into agentic commerce – the wave forcing brands to re-architect their websites and data layers for both human and AI agent audiences – and what that means for analytics teams. The episode closes on the AI analyst role itself: Eliot draws a direct parallel to how Clay created the GTM engineer out of rev ops, arguing the same elevation is coming for analysts – not replacement, but a shift to higher-order reasoning. His single best piece of advice for anyone navigating this moment: play with as many tools as you can.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant talks with June Dershewitz – a pioneering analytics leader who has been at the center of the data community for over two decades. June started her career in web analytics in 1999, co-founded Web Analytics Wednesdays (the industry gathering that launched a thousand local analytics communities), served as President of the Board of the Digital Analytics Association, and has led analytics and data governance teams at some of the largest media and entertainment platforms in the world. Now, as co-founder of InvestInData – a collective of 50+ Chief Data Officers and VPs of Data who angel invest in early-stage data startups – she sits at the unique intersection of practitioner, community builder, and investor.
Together they explore what 25 years of analytics evolution teaches you about the current AI transformation, how angel investing from the practitioner seat gives you a fundamentally different lens on which AI tools will actually matter, and why the community-building instinct that drove Web Analytics Wednesdays is more relevant now than ever as analysts figure out what the AI Analyst role actually looks like. June shares her perspective on scaling data teams through every major disruption – from the early days of web measurement through big data, real-time analytics, and now agentic AI – and what she’s learned about the human skills that no technology cycle has managed to automate away.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant talks with Tim Wilson – one of the most respected voices in digital analytics, widely known as the Quintessential Analyst (a title he’ll deny but absolutely deserves) and equally famous for climbing on a soapbox and delivering the kind of rants that somehow leave you smarter when he’s done. Tim has been working with digital data full-time since 2001, holding senior analytics roles at Search Discovery, Analytics Demystified, and across multiple agencies and Fortune 500 consultancies, and is widely known for his no-nonsense, clarity-first approach to getting business value out of data – earning him a reputation as one of the industry’s most beloved (and self-admittedly cranky) analytical thinkers.
Continuing the agentic e-commerce series, this episode goes upstream: before you instrument, before you build the data layer, how do you decide what to measure? Tim draws on over two decades of experience to argue that the agentic commerce shift – comparable in scale to mobile, Amazon, and GA4 combined – demands business clarity first, not more data collection. Together they explore why organizations keep repeating the same measurement mistakes across every technology disruption, how to use hypothesis testing and primary research to cut through the hype, and why the analyst’s real superpower is resisting the urge to solution before the business question is clear. The conversation also dives into the evolving skills analysts need now, from understanding LLMs to prompt and context engineering as the new SQL.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
Josh Silverbauer is Head of Analytics and CRO and Partner at From the Future, where he architects data layers and analytics implementations for brands navigating the messiest parts of measurement. He’s also the host of The Third Party Show – a comedy-musical talk show for digital marketing – and the creator of a rock opera about an alien named Cookie who loses his universe when Universal Analytics sunsets. In this episode, we dig into what happens when a purchase occurs without a browser session. OpenAI’s Instant Checkout means transactions can complete entirely server-side, with no JavaScript firing, no user journey, and no attribution – just an order appearing in your backend. Josh walks through where “ChatGPT / not set” shows up in your unassigned traffic, why server-side webhooks are currently the only way to count agent-driven purchases, and what the GA4 migration taught us about not dragging old assumptions into new architectures. We also explore why Microsoft Clarity is ahead of Google on AI traffic dashboards, and why building the agentic data layer feels less like a migration and more like a construction project with no blueprints.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
“We have so much trash in terms of websites that AI is now supposed to ingest and give us something meaningful. It’s not going to work.” Sani Manić has spent 15+ years optimizing websites for speed, accessibility, and technical health. A contributor to Search Engine Journal and WordPress Core, he’s now turned that expertise toward a new challenge: making the web work for AI agents. In this episode, we explore the emerging data layer beneath the agentic web – where users never see your brand’s colors or copy because an AI assistant completed the purchase for them. Sani and Katrin discuss why clean semantic HTML is the new competitive advantage, why you should never trust a tool built by someone who couldn’t do the job without AI, and the real security exposure of frontier models. Sani hosts the NoHacks Podcast, now subtitled “Optimizing the Web for AI Agents” and is building an auditing tool that shows how different LLMs and agents see, navigate, and transact on your website. For analysts and e-commerce teams, this episode makes a compelling case that GA4’s event-driven architecture and the BigQuery skills analysts built during migration may be their best preparation for tracking the agentic web, even if nobody planned it that way.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
“My mom called me about this cool thing called an LLM. I was like, I’ve been working on this for 10 years.” Yaniv Makover, CEO of Anyword, joins Avigad Oron, Head of Technology for Ask-Y, Two AI engineers with combined 40 years of experience dissect their ChatGPT moment and what it means for the future of work. We discuss the architecture decisions behind enterprise AI-from RAG and ranking to security nightmares where “your data goes into a super complex distributed system with multiple models.” They debate whether transformers will reign forever and agree on one thing: AI is pushing everyone toward senior-level integrative skills while routine tasks disappear.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
Jim Gianoglio has worn every hat in marketing-from photographer to creative director to SEO specialist to data scientist. Now he runs Cauzal Analytics, a boutique consultancy focused on marketing mix modeling and incrementality testing, and co-hosts MMM Hub, a community and newsletter for measurement practitioners. He also co-hosts the Measure Up podcast and built the session board app that digitizes Measure Camp schedules across the country.
In this episode, we explore the dual life of the modern analyst: using AI to power analytics workflows while also building AI-powered tools. Jim shares the “zero to 80% magic moment” of vibe coding with Replit, the frustration of context poisoning when LLMs get stuck in loops, and why he keeps re-testing tasks that failed months ago-because the models are always getting better. We dig into how analysts can future-proof their careers by listing every task they do, prioritizing by value, and systematically automating from the bottom up. Jim also reflects on how context engineering applies to marketing mix modeling and why understanding your business remains the skill AI can’t replace.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
What happens when AI meets bad data? Patrick Thompson has been building and using data tools for over 15 years. He started in growth at Atlassian, then co-founded Iteratively-a data quality platform acquired by Amplitude in 2021-where he became Director of Product. Patrick knows what it takes to build data layers, event pipelines, semantic layers, and quality monitoring systems. He knows what breaks when instrumentation goes wrong. In this episode, we explore how AI is lowering the floor for analysis while raising the bar for what makes analysts valuable. Patrick shares why natural language queries are changing user expectations overnight, why memory and context remain the hardest problems for AI to solve, and the difference between using AI as a tool versus a crutch. We dig into why gut-checking analysis is becoming a critical skill, and what Patrick tells people who ask if their kids should still study software engineering.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
A very special episode of Knowledge Distillation with Michael Helbling, co-founder and co-host of the Analytics Power Hour podcast and founder of Stacked Analytics. Michael has spent two decades building analytics programs as both consultant and practitioner, and he’s not one to sugarcoat anything. Together we break down why dropping an LLM on your data warehouse won’t magically produce insights, why semantic ambiguity kills most AI analytics initiatives, and what the streetlight effect means for AI’s ability to conduct real analysis. We get into the path forward – multi-agent architectures, semantic layers, knowledge graphs – while staying clear-eyed about the heavy lift required. Essential listening for anyone caught between CEO mandates to “AI everything” and the reality that this stuff doesn’t work like the LinkedIn posts suggest.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant speaks with Shomik Ghosh, a Silicon Valley VC who has lived through multiple tech hype cycles from both the investing and operating sides. Together they break down how hype forms, peaks, crashes, and ultimately reshapes roles – from DevOps to data science and now the rise of the AI analyst. Shomik explains why this generative AI wave is an enablement shift, spreading faster than any previous platform shift, and why embracing a learning mindset is now the most valuable career differentiator. They discuss which analyst skills will survive every hype cycle – deep understanding of data, critical thinking, and the ability to evaluate AI-generated logic. The episode also explores the future of work with AI agents and what it means to orchestrate virtual teams of autonomous tools. A must-listen for anyone navigating the new analytics landscape and trying to choose what to learn next.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this episode of Knowledge Distillation, Katrin Ribant talks with Mike Driscoll – co-founder and CEO of Rill Data and co-inventor of Apache Druid. Mike reflects on the early days of big data, the birth of real-time analytics, and what it meant to be an analyst in 2010. Together, they explore how today’s challenges have shifted from scale to maintaining analytical context across complex workflows. Mike discusses the rise of more technical, code-native analysts and how AI agents are transforming the way data teams operate. He shares why data engineering remains a superpower and how analysts can upscale to thrive as orchestrators of AI-driven processes. The episode offers a forward-looking view on the “AI analyst” skill set and the future of self-serve analytics.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
In this kickoff episode of Knowledge Distillation, host Katrin Ribant sits down with Martin Kihn, AI strategist at Salesforce and author of Agent Force. Together they explore how the role of the data analyst is shifting from manual data wrangling to orchestrating AI-powered workflows. Martin breaks down why understanding LLMs, prompt engineering, and AI-generated code is becoming essential for modern analysts. They discuss the misconceptions around AI replacing analysts and explain why critical thinking and business context matter more than ever. The episode highlights how AI accelerates hypothesis testing, improves visualization, and expands what analysts can achieve. A great introduction to the emerging skill set of the AI analyst.
All episodes on our website: www.ask-y.ai/knowledge-distillation-podcast
Learn more about ASK-Y: www.ask-y.ai
Comments
Download from Google Play
Download from App Store
United States