Docusign's CEO on the dangers of trusting AI to read, and write, your contracts

What It Covers

DocuSign CEO Alan Tiggeson explains how the company is expanding beyond electronic signatures into intelligent agreement management using AI. He discusses managing 7,000 employees, the liability risks of AI-powered contract summaries, why foundation models are becoming commoditized, and how DocuSign processes 150 million private agreements monthly to improve accuracy while maintaining trust in a $3 billion enterprise software business.

Key Questions Answered

• •Enterprise AI Cost Economics: Foundation model costs per token have dropped so dramatically that DocuSign now bundles AI features into standard subscriptions for mid-sized customers instead of charging separately. This commodity pricing dynamic forces model providers to differentiate through vertical integration—OpenAI pursuing consumer ads, Google leveraging cloud infrastructure, Anthropic focusing exclusively on enterprise coding tools—creating market instability for companies dependent on these models.
• •AI Accuracy Through Private Data: DocuSign's extraction accuracy dropped 15 percentage points when moving from public agreements to private corporate contracts. The company rebuilt accuracy by accumulating 150 million consented private agreements, adding tens of millions monthly. This proprietary dataset creates a competitive moat that public foundation models cannot replicate, demonstrating that domain-specific training data matters more than general model capabilities for specialized enterprise applications.
• •Digital Signature Identity Framework: Electronic signatures combine two distinct functions: identity verification through email delivery, IP tracing, and audit trails that hold up in court; and consent marking through any action indicating agreement. The signature appearance itself matters less than the verified identity database confirming who clicked agree. This reframes e-signature platforms as identity and consent databases rather than document tools, explaining DocuSign's resilience against acquisition attempts.
• •Agreement Workflow Inefficiency: Despite twenty years of electronic signatures, only 30 percent of agreements at large banks that have used DocuSign for over a decade are actually digitized and automated. Companies digitize high-value workflows first, leaving the majority of agreements in manual email-based processes. This represents massive expansion opportunity within existing customers before needing new customer acquisition, particularly through workflow automation and intelligent document preparation.
• •AI Liability Management Strategy: DocuSign delayed consumer-facing AI summaries for years despite having internal summarization tools, implementing legal disclaimers and graphical design to clarify the AI provides context, not legal advice. The company framed this as a moral obligation rather than just legal protection—consumers already paste agreements into ChatGPT, so providing summaries within a trusted platform with proper guardrails improves outcomes compared to uncontrolled external AI usage.
• •Product-Led Transformation Structure: Tiggeson shifted DocuSign from sales-centered to product-centered organization by reallocating investment from sales and marketing into engineering, growing the engineering team to 1,200 people. He implemented self-service capabilities modeled on Google's approach where billion-dollar advertisers place their own orders. The company now deploys new products in under twenty days and has 25,000 customers live on the AI platform within eighteen months of launch.