Podcast Summary: The Last Invention is AI
Episode Title: What VCs Are Looking For in AI Startups Today
Air Date: March 3, 2026
Host: Jaden Schaefer

Episode Overview

Host Jaden Schaefer explores what venture capitalists (VCs) are prioritizing—and avoiding—when investing in AI startups in 2026. The episode delves into the evolving criteria for funding, highlights industry opinions from noted investors, and discusses why depth, defensibility, and true automation have become more important than ever in the AI landscape. This analysis is driven by industry interviews, reports, and Schaefer’s own experience.

Key Discussion Points and Insights

The Shift in VC Investment Criteria (00:00–03:00)

• Past vs. Present:
  • In previous years, merely branding a project as “AI” could help secure funding, but that's changed.
  • “You can't just put AI on your pitch deck and get money.” (Jaden Schaefer, 02:00)
• Flood of AI Startups:
  • Nearly every startup now markets itself as an AI company, raising the bar for differentiation.
• Changing Expectations:
  • Investors are now highly selective—focusing on substance, not buzzwords.

What VCs Are Prioritizing in AI (03:00–09:00)

• According to Aaron Holiday (Managing Partner, 645 Ventures):
  • Top interest remains in AI-native infrastructure, vertical SaaS built on proprietary data, and platforms embedded into mission-critical workflows.
  • “AI that actually completes something” is the key focus, not just tools with a chat interface. (04:30)
• The New Standard:
  • Investors want “products that own something really essential”—tools that deliver end-to-end value autonomously, not just assist via AI.
  • Example: Instead of software that helps generate ideas for podcast titles, they want systems that automatically generate, schedule, and optimize entire podcast postings.

What VCs Are No Longer Interested In (09:00–13:00)

• Decreased Interest:
  • Thin workflow layers: Generic horizontal tools, surface-level analytics, and light product management tools can be quickly replicated by agents and large models.
  • Lack of “Data Moat”:
    • Abdul Abdirhan (F Prime): Vertical software without proprietary data moats is no longer compelling.
    • Having unique, hard-to-access data is a true differentiator.
• Direct Threat from Major AIs:
  • Startups offering slight AI improvements are being overtaken by foundation model providers (e.g., Anthropic launching “Anthropic for Legal” threatens tools like Harvey AI).

Depth and Defensibility Over UI and Automation (13:00–16:30)

• Igor Rabensky (Altal R Capital):
  • “If your differentiation mostly lives in UI and automation, that’s no longer enough. The barrier to entry is dropped, which makes building a real moat a lot harder.” (14:10)
• Advice for Founders:
  • Depth—ownership of unique workflows and granular understanding of a target vertical—is essential.
  • Massive codebases or clever UI are no longer meaningful moats.
  • Speed, focus, and adaptability now trump code volume.

Pricing Models and Consumption-Based Approaches (16:30–18:30)

• Subcriptions Are Out, Consumption Is In:
  • Traditional per-seat or flat subscription pricing is losing favor.
  • Consumption-based pricing (pay for what you actually use) is increasingly popular.
  • Example: AI Box's token-based model allows flexible, scalable usage.

Developer Tools & Task Execution (18:30–21:00)

• Jake Sapper (Emergence Capital):
  • Compares tools like Cursor (which owns workflow) versus Claude Code (which simply executes the task).
  • "Developers are choosing execution over process." (19:45)
  • VCs want products that get things done, not just manage processes.
• Implication:
  • Human–software “stickiness” is less valuable than in previous SaaS generations because AI agents now execute many tasks without manual human direction.
  • Software interfacing and integration tools (e.g., Zapier, Bubble) are losing their competitive edge as AIs mature in direct control and automation.

M&A and the Exception to the Rule (21:00–24:00)

• Calai Acquisition by MyFitnessPal:
  • Despite being a “thin AI wrapper,” Calai’s strong user growth (15M downloads), significant revenue ($30M+ annualized), and smart content marketing led to a major exit.
  • “If you have another angle… and you could get $30M annual recurring revenue, guess what, you’re going to get an acquisition.” (Jaden Schaefer, 22:00)
• Lesson:
  • Even simple tools can win big—but may not suit VC funding or be as defensible for long-term growth without strong moats.

What VCs Still Want (24:00–26:00)

• Lasting Themes:
  • Depth, domain expertise, proprietary data, and workflow ownership remain crucial.
  • Investors are reallocating toward startups with defensible assets, moving away from those easily copied or replaced by increasingly capable AI agents.
• “Becoming different isn’t just about kind of adding an automation, it’s about really owning something that agents can’t replace easily.” (Jaden Schaefer, 25:15)

Notable Quotes & Memorable Moments

• "You can't just put AI on your pitch deck and get money."
  — Jaden Schaefer (02:00)
• "What they're looking for is AI that actually completes something."
  — Aaron Holiday (04:30)
• "Vertical software without any proprietary data moats is no longer super compelling."
  — Abdul Abdirhan (11:20)
• "If your differentiation mostly lives in UI and automation, that's no longer enough. The barrier to entry is dropped, which makes building a real moat a lot harder."
  — Igor Rabensky (14:10)
• "Developers are choosing execution over process."
  — Jake Sapper (19:45)
• “SaaS companies struggling to raise capital are the ones that can easily be rebuilt.”
  — Igor Rabensky (21:30)
• "Becoming different isn’t just about kind of adding an automation, it’s about really owning something that agents can’t replace easily."
  — Jaden Schaefer (25:15)

Key Takeaways

• VCs in 2026 want depth, proprietary data, and defensible automation—not just AI labels.
• Shallow SaaS wrappers and “just another AI chat” tools are out; true workflow automation and unique data ownership are in.
• Consumption-based pricing models are ascendant.
• AI agents are eroding the moat of traditional workflow and integration tools.
• Simple, fast-growing AI apps can still have valuable exits, especially with growth-hacking and niche focus.
• To win funding, build something agents/foundational models can't easily copy or replace.

For listeners or founders, this episode delivers a practical guide to understanding how the VC landscape is shifting and what truly matters for fundraising and sustainable differentiation in the booming world of AI startups.