Episode Overview
In this episode of The Margin, Managing Director Igor Stenmark sits down with Grant Peterson, Chief Product Officer at Conga, to dissect the profound operational disruption and intense market hype surrounding Generative AI in Contract Lifecycle Management (CLM). Drawing on his extensive product engineering background, including his tenure as the driving force behind DocuSign’s e-signature product, Peterson provides a candid evaluation of AI capabilities in high-stakes legal and pricing environments.
While mainstream LLMs have introduced unprecedented text fluency, they also present a critical operational hazard: producing highly believable but structurally flawed output. This discussion strips away the veneer of standard vendor marketing to examine the technical realities of multi-pipeline AI architectures, the heavy transaction costs of running large models, and the looming legal and data leakage risks threatening corporate intellectual property.
Key Analytical Takeaways
- The “Product Propaganda” Hazard in Legal Tech: Why the 80% accuracy threshold of generative models presents a hidden risk profile for general counsels, creating superficially perfect contract drafts that harbor critical, high-liability inaccuracies.
- The Hybrid AI Strategy (Gen AI + Trained ML): A granular look at why the future of CLM relies on a multi-pipeline architecture, leveraging foundational LLMs for baseline public-domain clause extraction, followed by local, private machine learning models to map hyper-specific, confidential corporate terms without data leakage.
- De-Risking Implementation via Legacy Ingestion: How enterprise buyers can deploy Gen AI to dramatically compress implementation times, automating the extraction of core clause libraries and Best Alternative to Negotiated Agreement (BATNA) guardrails directly from legacy paper and digital repositories.
- The Silicon Layer as the New Level Playing Field: Why individual software vendors cannot claim proprietary “secret sauce” in core model development, and why a vendor’s true differentiation rests solely on prompt engineering, pipeline orchestration, and user-experience integration.
- The Looming “Napster Moment” for Enterprise IP: An objective evaluation of the systemic copyright and intellectual property risks associated with LLM training datasets, exploring how corporate buyers can insulate themselves from impending structural regulations.
- The Heavy Financial Floor of Transactional AI: Unpacking the underestimated computational costs of running full contract lifecycle processes through advanced LLMs, and the upcoming pricing adjustments buyers must navigate.
Featured Experts
Igor Stenmark | Managing Director, MGI Research
Igor brings his 30+ years of experience in entrepreneurial, strategic advisory, investment management, and executive roles in the technology industry to his clients. He serves as a strategic adviser to technology buyers, investors, boards, and management helping them make more informed decisions, enter new markets, optimize positioning, and build lasting value.
Grant Peterson | Chief Product Officer, Conga
A seasoned enterprise product leader and technical architect, Grant oversees business innovation and corporate technology strategy across high-stakes contract automation, document workflows, and agile billing solutions.
