Interview, Fireside Chat
Deal Velocity, Not Billable Hours: How Crosby Uses AI to Redefine Legal Contracting
- Lawyers and domain experts are expected to face increased demand for the skill of explaining AI-generated content, a capability identified as highly prized as routine tasks become automated.
- The legal services model is predicted to shift toward offloading undesirable weekly tasks to AI, creating a "magical experience" for professionals while introducing faster innovation cycles through unique feedback loops focused on review speed and word choice.
- Organizational structures modeled after law firms are anticipated to be adopted to establish safety parameters, ensure expert verification of quality, and align incentives by removing the billable hour, enabling precise pricing based on predicted negotiation cycles.
- Turnaround times are projected to drop from under an hour to mere minutes, with specific outcomes like term agreements forecasted to save a full week per negotiation by reducing exchange turns and allowing AI agents to simulate contract discussions.
- Risk allocation and terms such as governing law are expected to transition from subjective mental models to quantitative statistical guesses, utilizing client-specific context and fine-tuned models to achieve high accuracy in enterprise verticals.
- The market is forecast to see a "golden age" of legal innovation over the next five to ten years, with corporate law firm roles remaining relatively safe while a new net market emerges to fully automate individual legal services like child support and leases.
- Future workflows will likely replace roles similar to junior associates, senior associates, and junior partners, leading to senior associates managing "armies of agents" and specialized "AI-first" startups leveraging in-house domain expertise.
- Founders anticipate challenges with language subtleties and the overutilization of non-contractual data, necessitating domain-specific model training and the use of reinforcement learning to improve comment generation and explanation capabilities.
- Strategic plans include partitioning data to allow AI agents to negotiate against one another without exposing confidential information, creating auditable records and reducing the need for human touchpoints while meeting SLAs.
- Significant competitive advantages are expected from keeping domain experts and engineers in-house rather than outsourcing data, addressing the current lack of tuning in foundation models for contract law and specific verticals.
- Risks and barriers identified include the struggle of AI to grasp nuanced language changes, the potential danger of arrogance in refusing AI tools, and the historical difficulty of law firms teaching prompt engineering and iterative innovation effectively.