Overview of The Jaeden Schafer Podcast
Host Jaeden (Jaden) Schaefer breaks down what venture capitalists are funding in AI in 2026 — and, importantly, what they’re avoiding. Using quotes and themes from several VCs and industry examples, the episode argues that capital is moving toward AI products that own workflows, use proprietary data, and actually complete tasks (agents/execution), while pulling back from thin horizontal tools, UI-only differentiation, and shallow “AI-wrapped” SaaS.
Key takeaways
- VCs favor AI-native infrastructure, vertical SaaS with proprietary data moats, and systems that complete tasks (agents/systems of action).
- Investors are downgrading interest in generic horizontal tools, thin workflow layers, surface-level analytics, and UI/automation-only differentiation.
- Defensibility now often comes from proprietary data, deep integrations into mission‑critical workflows, and domain expertise — not large legacy codebases or per-seat pricing.
- Pricing models are shifting toward consumption/token-based models vs fixed per-seat subscriptions.
- Thin AI wrappers can still succeed commercially (growth/exit possible), but are less likely to draw large VC rounds unless they show extraordinary traction.
What VCs are looking for
Primary attributes
- AI that "actually completes something" — automation that performs end-to-end tasks rather than providing side‑help or chat prompts.
- Vertical, domain-specific products built on proprietary datasets (data moats).
- Deep embedding into mission-critical workflows — products that own the workflow and are essential to operations.
- Execution-first developer tools: tools that get work done for developers rather than merely orchestrating process.
Why these matter
- Agents that can execute tasks reduce the value of human-in-the-loop workflow stickiness; owning the human interface is less important than owning the task.
- Proprietary data and domain expertise create barriers to replication by big LLM providers or simple clones.
What VCs are avoiding
- Generic horizontal SaaS and thin workflow layers (e.g., basic PM tools, CRM clones).
- Surface-level analytics and UI/automation differentiation — if an LLM can replicate core value quickly, the product isn’t defensible.
- Large legacy codebases as a moat — speed, adaptability, and architecture matter more than “massive” code.
- Fixed per-seat pricing models are weaker compared to consumption/token billing for AI workloads.
Notable examples and industry signals
- Anthropic and other model providers launching verticalized offerings (e.g., "Anthropic for legal") can displace niche AI tools (example: Harvey for legal being supplanted by a large model’s vertical product).
- Acquisition example: a calorie-tracking AI app (Cal‑AI / similar) acquired by MyFitnessPal despite being a “thin wrapper” — success driven by immense user growth and revenue ($30M ARR cited) rather than VC-type defensibility.
- Tools like Cloud Code (execution-first) vs. Cursor (workflow ownership) illustrate developer preference for execution.
Implications for founders
- Prioritize building proprietary data assets and deep, real integrations into customer workflows.
- Design products that execute tasks autonomously (agents/systems of action) rather than offering chat-as-a-feature.
- Focus on domain expertise and workflow ownership from day one; shallow depth is a red flag to investors.
- Consider consumption-based pricing tied to token usage or actual work done rather than fixed seats.
- Don’t rely on UI/automation alone for moat — think about data, model training advantages, privileged access, or unique user behavior signals.
Actionable recommendations (for founders)
- Identify a mission-critical workflow you can own and map where an AI agent can complete the task end-to-end.
- Build or acquire proprietary datasets that competitors and public LLMs cannot easily reproduce.
- Architect for speed and adaptability: make it easy to iterate models and integrations.
- Test consumption-based pricing in pilot customers; align value to tokens/actions rather than seats.
- If you’re a horizontal or “thin” product, double down on growth channels or unique distribution (TikTok/social growth, integrations with large platforms) to be acquisition-worthy.
- Document domain expertise clearly: case studies, curated datasets, customer behaviors that form a defensible moat.
Selected quotes from the episode
- “AI that actually completes something.” — central investor criterion (emphasized by the host).
- “Shallow product depth is a really red flag.” — paraphrased from Igor Ryabensky.
- “The SaaS companies struggling to raise capital are the ones that can easily be rebuilt.” — investor summary used in episode.
Bottom line
Capital in 2026 flows to AI products that own real workflows, use proprietary data, and deliver autonomous execution. Founders building shallow, UI-first, or generic horizontal tools should either find defensible data/distribution advantages or expect lower VC interest — though commercial exits remain possible with strong growth.