DocuSign and Elastic supercharge generative contract and search solutions

At VentureBeat’s Transform 2024, Elastic CEO Ash Kulkarni and DocuSign CPO Dmitri Krakovsky shared insights on how generative AI is transforming enterprise search and contract management. Their discussion highlighted the growing importance of AI-powered search capabilities for businesses dealing with vast data volumes and complex contractual relationships.

Enhancing unstructured data analysis

Elastic’s approach to enterprise search has evolved significantly with the integration of generative AI. In May 2023, the company introduced the Elasticsearch Relevance Engine (ESRE), marking a pivotal shift in its search technology strategy. ESRE combines traditional keyword-based search methods with advanced vector search capabilities, enabling a more nuanced understanding of context and semantics in vast data repositories. This hybrid approach allows Elastic to offer its customers more sophisticated ways to retrieve relevant documents from their Elasticsearch stores, whether through vector search, text search using BM25 or a combination of techniques.

This year, Kulkarni expanded on the firm’s approach, highlighting the company’s progress in incorporating retrieval augmented generation (RAG) into its vector database technology: “We’re probably one of the most widely used vector databases out there, and we’ve implemented it.” Kulkarni said Elastic integrated these features into its database functionality, “so it benefits from all the work that we’ve done in Elasticsearch over the years.”

Kulkarni detailed the comprehensive capabilities now available: “So enterprise RAG capabilities, like permissions, faceted search, hybrid search, being able to do redirecting and use multiple search techniques like the BM25 basic search and so on with vector search, you get all that out of the box.”

Emphasizing Elastic’s commitment to flexibility and developer choice, Kulkarni added, “So in my opinion, it is super important to give developers all that choice and that functionality and give them openness on what models to use. That’s going to be part of our DNA.”

The rapid pace of AI development makes model choice crucial. Kulkarni noted, “Already, we are seeing customers split queries and send it to different multiple, different, large language models to try and get the best mix of accuracy and cost.”

AI-Driven contract management advancements

DocuSign is applying AI to transform contract management. Krakovsky outlined their vision: “We’re totally just scratching the surface in our space, having agents to help negotiate contracts.” This approach suggests a future where AI actively participates in contract negotiation processes.

The Intelligent Agreement Management (IAM) platform aims to convert static contract data into actionable insights.

IAM’s core components – Maestro, Navigator and App Center – work together to analyze contracts. The platform addresses a critical gap in enterprise digitization. As Krakovsky points out, “Contracts is this weird domain [where] most of the enterprises have been digitized for decades…  [but] contracts were sort of digitized, meaning that kind of PDFs are floating somewhere. But there’s no easy way to reason, to query, to understand them at scale.”

IAM transforms static PDFs into structured, analyzable data. Krakovsky illustrates its impact with a compelling example: “A customer had something like 70 contracts with an SI [system integrator]… They hire them to do a project here and another project there. And then they realize we’re spending hundreds of millions of dollars with that. And all of them have different terms because they all negotiated separately.”

By aggregating and systematically analyzing these contracts, Docusign’s approach enabled the customer to identify inconsistencies and opportunities, ultimately “shaving off over $100 million of spend.” This process, which once required manual review of hundreds of contracts, can now be largely automated, demonstrating IAM’s potential to transform contract management from a labor-intensive task to a strategic business function.

Navigating AI adoption challenges

Both executives emphasized responsible AI adoption. Krakovsky cautioned, “We have to move fast, but we have to move with caution, and make sure that we think through and answer questions that arise from using AI.” This approach is crucial when dealing with sensitive data like contracts.

DocuSign prioritizes data security and transparency. Krakovsky stated, “We’re just very cautious and very transparent. So, everything we do, we don’t hide stuff in terms of service and page number 325, you know, we make sure that if, even when we use customer data for training, we ask for implicit permission.”

Both executives stressed the importance of delivering end-to-end solutions rather than piecemeal offerings. Krakovsky emphasized: “We wanted to go after this because, time and time again, we talked to customers, and what we hear is that they want the problem solved end-to-end, not that them having to cobble together a set of stuff to solve it.”

Optimizing AI costs and resources

Cost optimization has emerged as a critical factor in AI adoption. Krakovsky noted, “Looking at how those things are put together, how we use them, how we call them, optimizing resources that we consume to stay part of the value on top of the site.” Efficient resource utilization is essential for sustainable AI scaling, especially for companies handling massive data volumes.

Kulkarni predicted changes in AI economics: “The cost of inference is going to keep coming down first and foremost, because of improvements in Hardware and Technology, like the kinds of chipsets you’re seeing, not just from Nvidia, but in other chip makers as well, but also competition in the area of large language models (LLMs).”

Looking ahead, both executives discussed AI’s expanding capabilities. Kulkarni spoke about multimodal AI models: “You can see a lot of models that are able to deal with multiple types of data and give you the right kinds of responses across fundamental modalities. And that’s really going to be more important.”

Krakovsky outlined specific applications in contract management: “Finding insights in contracts, ambiguity, compliance related stuff. A lot of this is findable and alertable and fixable. More or less automatically.”

Real-world AI implementation examples

The discussion included real-world applications of AI in enterprise settings. Kulkarni shared how Cisco used Elastic’s technology “to improve their customer support for internal use cases where the work that was being done by multiple engineers, they were able to automate it and free those engineers up for doing more interesting work.”

Kulkarni also provided an example from the financial sector: “A Fortune 100 bank is using Elastic to basically change entirely change the way their wealth managers talk with their clients.” AI-powered search tools are enabling more informed and personalized client interactions in wealth management by creating a “one-person Bloomberg terminal.”

From more intelligent search capabilities to AI-assisted contract negotiation, the potential applications are extensive. However, realizing this potential requires careful navigation of technical, ethical, and operational challenges. Issues of data privacy, model transparency, and cost-effective scaling remain at the forefront of discussions.