Two VCs On AI’s Possibilities — With Joe Marchese and Michael Mignano

Episode Summary

Executive Summary: Two VCs argue generative AI is a true platform shift, not a fad: it will automate boring workflows, reshape content creation and search, and become an assistive agent that takes actions. They see the biggest near-term opportunities in companies that already have strong products and can use AI as rocket fuel, while warning about copyright, deepfakes, misinformation, education, and labor disruption.

Main Topics: AI as an industry-wide transformation (Priority: 5/5): The guests reject the idea that generative AI is a crypto-like fad, arguing that it is already changing business processes across functions and industries by making human work faster and cheaper. Content creation and the collapse of old media economics (Priority: 5/5): AI is described as moving from directing attention to creating content itself, which could amplify already-existing platform dynamics and pressure ad-supported internet business models. Search, discovery, and disintermediation (Priority: 5/5): The discussion focuses on how AI answers questions directly, threatening search-based funnels and intermediaries like travel booking sites, while opening the door to new interfaces and business models. Assistive agents and action-taking systems (Priority: 5/5): A major theme is AI evolving from text generation into systems that can take actions through plugins and internet access, potentially eliminating steps in workflows and even whole software layers. Startup and VC opportunity (Priority: 4/5): Rather than backing pure AI models alone, the speakers favor startups with strong existing products that can use AI to improve efficiency, scale expertise, and create venture-sized outcomes. Risks, governance, and societal disruption (Priority: 5/5): They raise concerns about copyright/trademark law, deepfakes, misinformation, educational cheating, and unclear responsibility for AI-generated outputs, plus the broader issue of job displacement.

Key Arguments: Generative AI is not just incremental; it is pervasive and will affect most businesses, even if indirectly, because companies must figure out what role it plays in their workflows. AI is “fixing the boring stuff,” meaning legal, HR, writing, coding, and other micro-workflows that create immediate business value and efficiency gains. The biggest economic disruption may come from AI creating content and answers, not merely recommending content, which threatens search and advertising-dependent internet businesses. Plugins and live internet access turn chatbots into action-taking systems, enabling AI to move from information retrieval to transaction execution. The most promising startups are not pure model companies but existing products that can add AI to make creation or operations dramatically easier and more scalable. AI could produce broad productivity gains and GDP growth, but the gains may also reduce labor needs and require new social/economic frameworks such as UBI or a broader “purpose economy.” Risks include copyright/trademark uncertainty, the legality of training data, AI-generated bots built from a person’s work, and deepfakes that may fragment shared reality. Educational use is double-edged: AI can help creativity and learning, but it can also become a shortcut that prevents people from developing core skills themselves.

Data Points: Model training cutoff: September 2021 - Mike says GPT-4/ChatGPT was trained on a corpus of data up to around this point and was initially not live on the internet. GPT version referenced: GPT-3.5 - The host notes ChatGPT was an incremental step built on existing technology, specifically GPT-3.5 inside a chatbot interface. Product categories highlighted by the host: 3 - The introduction flags three areas to watch: content creation and entertainment, search and discovery, and assistive-agent behavior. Timeframe mentioned: 5-6 years - Mike refers to thinking back several years to earlier voice-assistant and command-bar efforts that did not fully solve interaction problems. Scale reference: hundreds of billions of dollars in market cap - Joe uses online travel agencies like Priceline, TripAdvisor, Expedia as an example of a vertical vulnerable to AI disintermediation. Company adoption lens: 1 engineer could do the work of 10 - Mike argues AI tools like Copilot could massively amplify engineering output, changing the definition of hard work. Surveyed business impact: every business - Joe says AI will force every business to determine what role it plays, likening it to cloud adoption rather than a standalone AI industry.

Pivotal Quotes: "AI is fixing the boring stuff." — Joe Marchese: Joe’s core thesis for why generative AI is real and durable: it improves mundane workflows like legal and HR rather than trying to remake governance or money. "AI isn't going to replace humans, but humans who use AI will replace humans who don't use AI." — Mike McNano (attributed as a common line discussed on the show): Used to frame AI as an adoption advantage for workers and companies that integrate it quickly. "When you ask ChatGPT, it doesn't go search the web and then give you a reference. It creates an answer for you." — Joe Marchese: Joe explains why AI changes search and advertising by generating content directly instead of merely pointing to it.

Implications: AI may become a default layer across products and workflows, reshaping search, media, software, and services. Winners will pair strong products with AI; losers may be middlemen, ad-dependent platforms, and organizations slow to adapt.