20VC: How Model Performance is Plateauing | Two Key Rules for Effective Deal-Making | Company Building Lessons from Keith Rabois, Brian Halligan and Pat Grady | Why Enterprise AI Adoption is Years Off with Harvey CEO Winston Weinberg

What It Covers

Harvey CEO Winston Weinberg discusses scaling to $190M ARR in legal AI, achieving 74% DAU/MAU with multi-product users, navigating $8B valuation expectations, and why enterprise AI adoption remains three to five years away despite current capabilities.

Key Questions Answered

• •Infrastructure hiring strategy: Allocate 40% of engineering resources to senior infrastructure engineers from companies like Databricks, not just front-end developers. Many AI companies over-hire front-end talent for demos but lack backend architecture to support hundreds of thousands of active users at scale.
• •Fundraising execution: Start fundraising six months early by bringing in small investors with information rights. Let them track progress against predictions over multiple quarters to build trust. This enables twelve-hour fundraising processes with targeted partners versus optimizing for maximum valuation through competitive processes.
• •Deal-making principles: Listen more than speaking in negotiations, and know when not to negotiate at all. When you understand value better than others, abandon traditional back-and-forth tactics and immediately secure the critical element you need, even if advisors object to financial terms or other conditions.
• •Gross retention focus: Prioritize GRR over net new ARR growth when scaling past $100M. Companies that chase land-grab metrics without infrastructure to support customers face mass churn. Post-sales investment matters more than pre-sales as single customers can scale from $1M to $100M annual spend with proper retention.
• •Product velocity management: Transition between three phases: product-market fit, company-market fit, then back to product-market fit. Each quarter's product planning should feel painful, like a breakup, requiring saying no to genuinely good ideas. Prioritization means rejecting features that hurt to cut, not just eliminating obvious non-priorities.