The Aperiam Podcast

Daniel Saedi, Co-Founder and Chief Strategy Officer of Minerva, explains how two former quants are building an AI-powered unified media platform for brands. Daniel joins Corey Ferengul and Joe Zawadzki to discuss bringing Bridgewater and Citadel discipline to marketing, Minerva's 250 million-record consumer file spanning consumer and professional data, the data-leakage problem that undermines marketing models, fast decision models crossing over from finance, background agents that unify customer data without a chat box, the shift from selling data to selling AI, and why B2B marketers remain underinvested in digital channels.

Erik Hovanec, former Chief Strategy Officer of Magnite, on how header bidding pushed Rubicon Project into eight acquisitions and why integration is harder than the deal. Erik joins Corey Ferengul and Joe Zawadzki to explain the chief strategy officer role, how to tell a pan fire from a house fire, and how to build consensus by leading from behind. He walks through the SpotX and SpringServe deals, his view of ad tech as an intermediary business that wins on transparency and scale, and why extractive margins invite competition. For founders, he covers what kills deals, why acquirers buy execution rather than vision, and what AI does and does not change about pricing, margins, and cap tables.

Adam Epstein, co-founder and CEO of Gigi, on building an AI media manager for the Amazon DSP and why agencies now hire software rather than buy it. Adam explains how Amazon's ad business grew from about $5 billion to almost $80 billion, why only an estimated 1,000 people globally have four-plus years of Amazon DSP experience, and how Amazon is turning categories like auto into pseudo-endemic ones. Corey Ferengul and Joe Zawadzki dig into Amazon's API-first partner model, Gigi's "learned context" contract term that lets customers own the IP their agent builds, why agents should go deep in one channel before going wide, and the road to outcome-based pricing. 

Mike Khristo, founder and CEO of Layers, joins Corey Ferengul and Joe Zawadzki on why app distribution, not development, is now the hard part. He spent six and a half years building ad products at Snapchat before starting Layers, an app studio that became growth infrastructure after he shipped his own app and watched the download counter sit at zero. They cover what Layers found analyzing 13 trillion seconds of TikTok video across 23 million posts, how account warming works, the new Layers Growth MCP and why agents rejected simple campaign tools in favor of research, web-to-app campaigns and web payments after the Epic ruling, and the case for the first mobile measurement agent. Layers is an Aperiam Ventures portfolio company. 

Momentiv co-founder Sandro Catanzaro on rebuilding the CTV ad server from scratch, and why AI made a long-unfundable idea buildable. Sandro, previously a co-founder of DataXu with stints at Roku, Pathmatics and The Trade Desk, joins Corey Ferengul and Joe Zawadzki to discuss the 18-hour prototype that started the company, why development is no longer the bottleneck, and how running two ad servers in parallel removes the migration risk that has protected incumbents for years. Also covered: identity and forecasting native to the ad server, agentic buying without publisher data leakage, scene-level contextual targeting in CTV, a thin publisher-only fee, and the coming arms race between buyer and seller agents. 

Iván Markman joined Mediaocean in May 2026 as its first President and COO — a role that didn't exist before he took it. He joins Corey Ferengul and Joe Zawadzki to explain what Bill Wise hired him to fix.

Iván traces a path from Overture and Yahoo through MarketShare and Neustar to running search, ad tech, and Edgecast at Verizon Media, where he helped turn the business around ahead of the Apollo sale. He describes Mediaocean as two halves of the same problem: Prisma, the financial backbone moving a couple hundred billion dollars of media and reconciling it to the penny, and the Flashtalking, Innovid, and 4C side serving roughly a trillion impressions a year across social, CTV, and everything between. The company buys no media and sells no media — Iván compares the position to Visa.

The conversation turns to what actually makes agentic advertising work. Iván argues the scarce asset isn't the model, it's deterministic data and deterministic pathways — and that Mediaocean owns a second data set nobody else has: not the client's campaign data, but the data about how the money and the impressions actually moved. On MCP, he notes the important letter is the C.

He also covers the integration reality behind the marketing, why he asks every founder for their "investor scorecard" rather than a pitch, Mediaocean AI Ventures, and where the next dollar goes: building.

Corey and Joe talk with Matt Byrne of OAREX, a lender built specifically for ad tech financing, about why long payment terms create a compounding working capital problem across the entire media supply chain. They cover the "ad tech financing tax," why traditional banks avoid the space, how OAREX uses direct DSP and SSP data integrations to underwrite credit banks can't, the lasting impact of the SVB collapse, and why more founders and investors are starting to treat debt as core infrastructure, not a last resort.

Corey and Joe talk with Brian Wiener — a 30-year digital veteran behind 360i and Profitero — about why he started 37Arc, a "workflow intelligence" company helping CMOs rewire marketing and sales for the AI era. They dig into why most AI pilots fail, how 37Arc uses a voice AI agent to interview entire marketing departments at scale, and why chasing shiny objects beats companies that actually map their workflows first. 

Corey and Joe talk with Jonathan Kroopf — who helped build Creator IQ and later led creator strategy at TikTok — about founding Devotion with co-founder Cami Tellas. They cover why influence is fragmenting from lifestyle influencers toward everyday people with day jobs, how Devotion helps brands scale from 20-30 macro creator posts a month to hundreds, and why building a real creator community beats transactional, campaign-by-campaign influencer relationships. 

Performance marketing used to be about who could do the most manual work across the most platforms. Anna Stepura, co-founder and CEO of Adlyse, says that era is over — and the teams still operating that way are already falling behind.

In this episode, Anna joins Corey Ferengul and Joe Zawadzki to break down why she built Adlyse as an agentic operating system sitting above Google, Meta, TikTok, OpenAI, and every other channel — not inside any one of them. She unpacks why platform-native AI is inherently biased toward getting you to spend more, why it now takes 21 cross-platform touchpoints to convert a customer (up from seven), and how Adlyse runs 100 different hypotheses before making a single optimization recommendation.

They get into the creative remixing loop, the 24-day trust arc from human-in-the-loop to autopilot, agent-to-agent advertising, OpenAI ads as an emerging channel, and why the platforms themselves are actively opening their APIs rather than closing them — because agentic buying is the future they're building toward too.