Episode Summary
Executive Summary: The episode is a rapid-fire demo session on how AI is reshaping startups, travel, newsletters, image generation, search, journalism, and website building. Jason and Sundeep argue that AI is moving fast enough to create real product differentiation, but also warn startups to build proprietary data, curation, and customer relationships before incumbents like Google, OpenAI, or Apple absorb the use case.
Main Topics: AI as an “elevator” toward AGI (Priority: 5/5): The hosts frame current AI progress as a dramatic upward trajectory toward AGI, emphasizing how quickly capabilities are improving and how each month reveals new leaps. Travel planning and vertical AI products (Priority: 5/5): Copilot to Trip is used as the flagship example of a startup adding value on top of LLMs by combining itinerary generation, mapping, and curated experiences. Fine-tuning, proprietary data, and platform risk (Priority: 5/5): They discuss OpenAI fine-tuning APIs, the value of owning custom datasets, and the risk of building on platforms that may eventually copy successful use cases. AI for newsletters, curation, and summarization (Priority: 4/5): Hoppy Copy is shown as an AI-assisted newsletter builder, and the conversation expands into how AI could automate summarization while pushing journalists toward original reporting. Multimodal AI and visual understanding (Priority: 4/5): Google Bard’s ability to explain memes and screenshots is highlighted as a major step in multimodal AI, with implications for humor, analysis, and writer workflows. Voice-first, conversational commerce and personalization (Priority: 4/5): The hosts argue that voice interaction will become the natural interface for shopping, planning, and collaboration, especially as memory and personalization improve. AI website builders and startup acceleration (Priority: 3/5): 10Web is showcased as a practical AI website builder, illustrating how founders can launch much faster by using AI-generated templates instead of starting from scratch.
Key Arguments: AI startups need more than wrappers; they must add curation, structured workflows, and proprietary datasets to stay defensible. Travel is an especially strong AI vertical because it is monetizable, transactional, and benefits from itinerary-level personalization. OpenAI fine-tuning can help startups preserve product differentiation, but building on a platform always carries copy-risk if the platform learns from user behavior. Journalism’s commodity layer is summarization, and AI will likely automate that layer, forcing writers to return to original reporting and real interviews. Multimodal AI can already explain jokes, graphs, and memes, which suggests future systems will understand context well enough to generate better content. Voice interfaces will be more useful than typing for many tasks because they let the AI interview the user, refine intent, and personalize outcomes in real time. AI website builders and other no-code tools lower the barrier to launching businesses, allowing founders to focus on the core product rather than infrastructure.
Data Points: ChatGPT Plus price: $20/month - Used as an example of a considered consumer purchase, not mass-market pricing. Startups.tech referral program: $? / featured startup promotion - Ad segment for Dot Tech Domains featuring startups.tech and DealMaker.tech. Inside.com analyst salary: $70,000/year - Mentioned as pay for newsletter analysts who curate and summarize news. Business Insider starting salary: $50,000–$55,000/year - Used as a comparison point for newsletter analyst compensation. AI summarization quality estimate: 60%–70% as good - Jason estimates raw ChatGPT summaries are roughly this far from inside.com’s human summaries. Training-set rough baseline: 100 examples - Sundeep suggests about a hundred original article-summary pairs as a starting point for fine-tuning. Potential automated performance after fine-tuning: 95% - Sundeep estimates a fine-tuned model could approach human-quality output quickly. 10Web build time: less than 30 seconds - The website builder generated a first-pass corporate site template very quickly. All In Summit planning timeline: 70 days away - Used while discussing AI-assisted event planning and project coordination. Threads surge in downloads: 30 million on July 5 - Bard’s multimodal explanation referenced a graph showing a spike in downloads.
Pivotal Quotes: "We're on the elevator going up." — Jason Calacanis: He uses this metaphor to describe the rapid climb of AI capabilities toward AGI. "I think this is the type of use case that folks should go after, right? That it's not just a straight wrapper around OpenAI." — Sundeep Madra: Discussing what makes an AI startup defensible in travel and other verticals. "The thing about journalism today is like I got the who, what, when, where, why from tweets." — Jason Calacanis: Explaining how AI may automate commodity reporting and shift value toward original reporting.
Implications: AI is moving from novelty to infrastructure. Startups that add workflow, memory, curation, and proprietary data can still win, but only if they move fast and avoid becoming easily copied feature layers.
About This Week in Startups
Jason Calacanis covers startups, tech, markets, media, and all the hottest topics in business and technology. He also interviews the world’s greatest founders, operators, investors, and innovators.