Key Takeaways
- Harvey demonstrated unprecedented scale, achieving $190M ARR, 500 employees, and over 1,000 customers.
- Founders' personal routines and daily habits are critical for managing stress and making strategic decisions.
- Building deep trust with a few key investors ensures more predictable and efficient fundraising processes.
- AI application companies face competitive pressure from rapid product development by major model providers like OpenAI.
- Enterprise AI adoption is slow due to complex integration into workflows, not a lack of AI capabilities.
- Prioritizing infrastructure and Gross Revenue Retention is essential for AI companies' long-term scaling and customer commitments.
- AI will likely expand the professional services market, creating new demand areas like AI risk and international expansion.
- Hiring in Europe requires longer-term strategic planning due to extended notice periods and hiring timelines.
- Effective deal-making involves strategic listening, understanding unique value, and knowing when not to negotiate.
Deep Dive
- The guest runs a mile daily to manage stress and improve decision-making while building a company.
- He consumes a liter of water upon waking, crediting it for immediate hydration and a sense of accomplishment.
- The guest notes a habit of checking Slack every 15 minutes, which aids early-stage decision-making but can hinder focus at scale.
- The host wakes around 4:00 to 4:30 a.m. East Coast time to focus on strategic thinking before daily communications.
- A 20-25x multiple on end-of-year revenue is considered reasonable, while a 100x multiple suggests a Series A round.
- Harvey's Series C round at a $1.5 billion valuation felt most uncomfortable due to lower revenue at the time.
- The guest focuses on building trust with a few key investors over time, aiming for preemptive or quick closes.
- Predictability and consistently hitting plans are crucial; few of their 170 investments consistently hit plan.
- The primary threat to application layer AI companies is the rapid pace of product development by major model providers like OpenAI and Anthropic.
- Maintaining a significant product delta is necessary to sustain a competitive advantage and avoid being overtaken.
- Model routing is use-case dependent, with the priority being the best-performing model, even if traffic shifts to a competitor's model.
- While consumer AI models plateau, enterprise AI needs better integration across systems, with CodeGen capabilities predicted to increase within 12 months.
- Enterprise AI adoption faces a three to five-year timeline for significant productivity gains due to complex workflow integration, not AI capabilities.
- Vertical AI solutions like Harvey see cross-departmental adoption beyond initial focus, such as legal teams interacting with compliance and HR.
- Building enterprise-ready security and permissioning systems, especially for bank customers, required nearly a year of development.
- Many AI application companies prioritize front-end engineers, leading to impressive demos but potential underinvestment in scalable architecture.
- AI tools like Harvey can help law firms win new business by enabling custom solutions, leading to significant ROI.
- Consumption-based pricing is emerging for AI platforms in professional services, with AI tools already being expensed as technology budgets.
- Harvey's revenue is split 40% from in-house corporate legal teams and 60% from law firms.
- AI is expected to increase overall work in professional services, creating new demand areas like AI risk and international expansion, rather than shrinking the market.
- Hiring in Europe involves significantly longer processes than in the US, primarily due to extended notice periods.
- This necessitates a longer-term strategic approach for establishing offices and building teams in Europe.
- The guest notes that lawyers globally are disciplined and hardworking, debunking the trope of Europeans not working as hard.
- An earlier delay in European investment was attributed to bandwidth constraints during rapid scaling, including onboarding a 4,000-person client with a four-person team.
- A VC prediction from late 2022 suggested many VCs would rely on resumes and established logos due to AI chaos, creating a disconnect with AI research.
- To identify top AI talent, the guest advises asking respected researchers in the merit-based and tightly knit community.
- AI researchers are motivated by solving difficult problems and may leave companies if the direction shifts away from their interests.
- The demand for AI talent is very high, emphasizing its renewed importance for developing differentiated products.
- The guest cold-emailed OpenAI's Sam Altman and Jason Kwan in summer 2022, presenting evidence of GPT-3's capabilities in legal contexts.
- This outreach led to a meeting and subsequent investment from OpenAI for Harvey.
- The guest pitched OpenAI's C-suite on July 4th, 2022, and secured multiple term sheets within 48 hours for their Series A.
- He contrasted his initial unfamiliarity with prominent VCs during seed and Series A rounds with his current knowledge.