What was Ironclad's first product before advanced AI?

Before advanced AI, Ironclad used an email alias (admin@ironclad.ai) where users could initiate transactions like NDAs. This system would then manage the process, collect information, and send documents for signature, with the founder manually handling the AI aspects initially.

How has Ironclad integrated modern AI models into its platform?

Ironclad realized workflows were a better way to create contracts accurately. Since around 2018-2019, with the advent of transformer models, they've focused on integrating AI for data extraction from existing documents and enhancing their contract lifecycle management (CLM) product.

Why are attorneys open to AI when the popular narrative suggests resistance?

Ironclad finds attorneys amenable to AI, contrary to popular belief. This is attributed to a decade of building attorney trust, allowing for more dialogue and adoption within their user base, potentially differing from broader industry trends.

What is Ironclad's strategy for developing AI technology versus applying it?

Ironclad believes verticalized application of AI is strategic for unique development, while foundational capabilities should leverage state-of-the-art models from providers like OpenAI, Google, and Anthropic, evaluating each model's strengths for diverse use cases.

Why has enterprise AI adoption in the legal field slowed?

The slowdown is less about technology features and more about buyer sophistication and comfort with data usage in AI. Concerns about data security and privacy, and the lack of established trust with newer vendors, hinder adoption, especially for startups.

What has surprised Jason Boehmig the most in the last 18 months regarding AI?

The biggest surprise has been the sheer pace of change in AI technology. Nine months ago feels like a different world, and this rapid evolution has provided immense energy and new learning opportunities for founders.

Jason Boehmig, CEO of Ironclad on Balancing Risk, Innovation, and AI Opportunity in the Legal Field

AssemblyAI

Science & Technology4 min read30 min video

Oct 2, 2024|767 views|14

Save to Pod

Key Moments

TL;DR

Ironclad CEO discusses AI in law, balancing risk and innovation, and building trust.

Key Insights

Ironclad's origin story stems from a lawyer's attempt to automate legal workflows, leading to the realization that industry-specific software was lacking.

The company’s early strategy focused on building robust workflows for contract creation and management, ensuring 100% accuracy before fully integrating advanced AI.

Ironclad prioritizes customer trust and rigorous testing, distinguishing between internal risk tolerance and the risk customers experience with their product.

Their technology strategy leverages foundational AI models from various providers while developing proprietary verticalized applications for unique customer value.

Enterprise AI adoption in legal is hampered by buyer sophistication and data security concerns, with established players like Ironclad having an advantage.

The rapid pace of AI development has been the most surprising aspect of the last 18 months, injecting energy and renewed focus into the industry.

FROM LAWYER TO INNOVATOR: THE FOUNDING OF IRONCLAD

Jason Boehmig, CEO of Ironclad, shares his entrepreneurial journey that began as a corporate attorney. Frustrated by the manual and archaic nature of legal document creation and management, he started automating his own workflows. Recognizing a significant gap in the market for modern, efficient software tailored to lawyers, he saw an opportunity to build what didn't exist. This realization, compounded by the lack of existing solutions from established software companies, fueled his decision to quit his job and embark on building Ironclad, initially with the goal of becoming a more technologically adept attorney, which ultimately evolved into founding a company.

EARLY STRATEGIES AND THE EVOLUTION OF AI INTEGRATION

Ironclad's initial approach focused on creating highly effective workflows for contract creation and management, aiming for 100% accuracy. This was crucial because errors in legal contracts are unacceptable. The company began with innovative methods like an email alias acting as a rudimentary AI assistant for transactions. As AI technology advanced, particularly with the advent of transformer models around 2017-2018, Ironclad began strategically layering AI capabilities into their platform, starting with data extraction from existing documents. This allowed them to generate valuable structured data from a large volume of contracts, complementing their existing workflow-driven approach.

BALANCING RISK, INNOVATION, AND CUSTOMER TRUST

Boehmig emphasizes a clear distinction between a company's internal risk tolerance and the experience provided to customers. While comfortable taking significant risks internally as a founder-led company, he insists on ensuring customers face no risk when using Ironclad's products. This principle guides their rigorous testing processes, especially for standalone AI products, explaining their seemingly slower public launch of certain features compared to industry noise. Winning and maintaining attorney trust has been paramount, allowing for more open dialogue and user feedback, which in turn fuels product development and ideation.

STRATEGIC TECHNOLOGY DEVELOPMENT: PROPRIETARY VS. FOUNDATIONAL AI

Ironcladviews its technology strategy as a balance between developing proprietary, verticalized applications and leveraging foundational AI models. They believe the unique application of AI to specific legal use cases, like contract language recommendation, is where their unique advantage lies and requires in-house development. However, for foundational capabilities, they aim to stay at the forefront by evaluating and integrating state-of-the-art models from providers like OpenAI, Google, and Anthropic, using a multi-model approach to best suit diverse use cases ranging from negotiation to data extraction. This allows them to adapt quickly without becoming overly reliant on a single provider.

NAVIGATING THE ENTERPRISE AI ADOPTION LANDSCAPE

Despite the buzz around AI, enterprise adoption in the legal sector remains low, often characterized by single-digit percentages in production. Boehmig attributes this not to a lack of technological features but to the sophistication of buyers and their comfort levels with data usage. Concerns around data security, especially regarding customer data being used for training foundational models, are significant hurdles. Ironclad, as an established scaled company with a strong focus on data protection and a sophisticated contract-based narrative, is well-positioned to address these concerns, unlike newer startups that may struggle to gain the necessary trust from legal teams.

THE ACCELERATED PACE OF CHANGE AND FUTURE OUTLOOK

The most surprising aspect of the past 18 months for Boehmig has been the unprecedented pace of AI development, making recent events feel like a decade ago. This rapid evolution has injected immense energy not only into the industry but also into founders like himself, fostering a constant drive to absorb new information and collaborate with peers. Despite the challenges of long-term founder journeys, the current AI boom provides a significant boost, encouraging continuous learning and innovation. This dynamism suggests that while the market is still catching up to the reality of AI's potential, the trajectory is clear.

Mentioned in This Episode

●Software & Apps

●Companies

●Organizations

●Concepts

Navigating AI Adoption in Legal Tech: Key Strategies

Practical takeaways from this episode

Do This

Focus on specific use cases within your product for AI integration.

Build and maintain trust with users, especially in fields like law.

Adopt a balanced technology strategy: develop unique vertical applications and leverage state-of-the-art foundational models.

Prioritize rigorous testing and a high bar for customer-facing AI features.

For incumbents, integrate AI across existing products while maintaining focus on the core offering.

Consider a multi-product strategy with performing and incubation zones for new ideas.

Focus on the delta in performance and user value rather than solely the 'how' of AI achievement.

Educate customers and address data security concerns to foster enterprise adoption.

Stay agile and adapt technology strategy rapidly to keep pace with AI advancements.

Embrace new information and founder-to-founder insights to stay energized.

Avoid This

Do not expect AI to be ready for market universally in nascent stages.

Avoid being lured into solely developing everything in-house if foundational models offer significant advantages.

Do not assume a custom-trained model provides a sustainable long-term advantage over staying on main branches.

Refrain from positioning AI as a complete replacement for legal professionals if the technology is not yet ready.

Do not let new AI product launches cannibalize existing successful product lines without careful consideration.

Avoid complacency; continuously seek to disrupt your own products and services.

Do not neglect the 'why' behind AI adoption – focus on generating tangible user value.

Do not underestimate the importance of building trust with users, especially in regulated industries.

Avoid getting stuck on outdated technology versions when newer, better models are available.

Common Questions

Jason Boehmig, an attorney, started Ironclad by automating his own legal workflows, realizing there was a lack of modern software for lawyers. The initial vision was to standardize and automate parts of legal practice, leading to the development of contract management solutions.

Topics

AI In Legal Legal Technology Contract Lifecycle Management Founding Story

Mentioned in this video

conceptRAG

Retrieval-Augmented Generation, a technique used to supplement foundational models and prevent hallucination, particularly useful for applications involving specific datasets like case law.

softwareGPT-4o

The latest advanced model from OpenAI, offering benefits over GPT-3 and considered superior to forked models, especially when combined with techniques like RAG for specific applications like legal data extraction.

organizationBain & Company

Mentioned for a report indicating that enterprise AI adoption in the legal profession has decreased year-over-year, highlighting a gap between perceived potential and production implementation.

companyIronclad

A legal technology company focused on contract lifecycle management, which has successfully integrated AI into its product offerings and achieved significant ARR growth.

toolMicrosoft Word

toolLlama 3

toolExcel