Knowable, a legal technology company specializing in helping organizations bring order and organization to their executed agreements, has announced Ask Knowable, a suite of generative AI-powered tools aimed at transforming how legal teams interact with and understand what is in their contracts. Released today as a commercial preview and set to launch for general availability […]
Knowable, a legal technology company specializing in helping organizations bring order and organization to their executed agreements, has announced Ask Knowable, a suite of generative AI-powered tools aimed at transforming how legal teams interact with and understand what is in their contracts.
Released today as a commercial preview and set to launch for general availability in March 2025, the feature marks a significant step forward in leveraging large language models to address the complexities of contract management, the company says.
“This suite is all about making contract data work for people, not the other way around,” Nik Reed, who became CEO of Knowable earlier this month, told me during a preview demonstration of the new features.
“We do this by combining new and older AI – utilizing gen AI’s unique conversational strengths, all the while grounding responses in highly accurate data and our proprietary contract family structure, which we create with traditional ‘ML’ and human-in-the-loop QC.”
Key features of the Ask Knowable suite include:
- AI Contract Search. Locate contracts in seconds using conversational prompts without complex filters or data modeling expertise.
- AI Contract Summaries. Get automatic summaries, not just of individual agreements, but also of entire contract families, reflecting how terms change as contracts grow and evolve and, most importantly, which terms are controlling right now.
- AI Chat: Get answers to questions about specific clauses, key terms and obligations by engaging directly with advanced gen AI that can understand the evolution of contract terms across interrelated agreements.
All In the Family
In our demonstration, Reed started by outlining the problems companies face in managing their contracts and Knowable’s unique approach to addressing those problems.
The problem starts with the fact that contracts contain mission-critical information and that companies face a constant flow of questions pertaining to their obligations and entitlements under those contracts.
But because contracts are typically poorly organized and difficult to access and analyze, answering those questions can be an odyssey, especially in organizations managing thousands of contracts.
Knowable organizes contracts by families and its AI allows users to ask questions about an entire family.
That inefficiency can have real cost implications, Reed said, as was revealed by a study Knowable conducted with one of its enterprise clients. The client performed over 73,000 contract-related searches annually, incurring approximately $10 million in costs. In one notable instance, the client, using Knowable’s system, uncovered a significant billing error — where a monthly fee was mistakenly charged as the annual cost. Simply correcting this mistake more than recouped the cost of the Knowable tool, he said.
While recent years have seen proliferation of all-in-one contract lifecycle management (CLM) systems, those systems are not up to this challenge, Reed believes. While they can be good for creating and negotiating new contracts to sign, they are not very good at managing executed contracts.
Rather, Reed says that the proper management of executed agreements — those that have been signed and are in effect — requires two non-negotiables.
One is data quality – the need for contract data to be highly accurate. While many organizations and vendors operate under the fallacy that it is easy to get this right, the fact is that even seemingly easy data points can be hard to pin down.
Take the example of a contract’s effective date. Just 21% of contracts clearly and explicitly state this as a date. Others might define it by the date of execution, or with reference to some external date or with reference to a separate agreement, or even based on the occurrence of some condition or event.
“When lawyers get crafty, it is hard for AI to understand, and lawyers get crafty all the time,” Reed said.
Knowable believes the standard of accuracy required for contract data is a minimum of 98%, Reed said, and that is the standard it promises to deliver to its customers.
The second non-negotiable for managing contracts, Reed said, is that families matter. For just a single counterparty, there may be multiple contracts that share aspects of an even-larger number of agreements (such as NDAs or DPAs). That means that the only way to truly understand a contract is to map it by its families.
“Agreements impact each other, and contract families have their own order of operations,” he said.
Reed said that Knowable has spent years developing a system tailored specifically to executed agreements, at the heart of which are those two non-negotiable pillars:
- Family mapping. Agreements are grouped into “families” that include the parent agreement and all related documents, such as amendments and addenda, creating a clear view of their interdependencies.
- Clean and accurate data. Knowable applies rigorous data extraction and quality control to create a set of “essentials data,” including attributes like effective dates, terms, and parties. All contracts are deduplicated, error-checked, and organized into a single repository, ensuring a consistent baseline for analysis.
Those non-negotiables drive Knowable’s approach to contract management, and set the stage for why the company believes its new AI features are unlike anything else on the market.
“If you don’t have families and you don’t have the accurate data in the first place, you can’t start to solve this problem,” Reed said.
Ask Knowable
Ask Knowable builds on this foundation, incorporating generative AI and large language models (LLMs) to enhance the platform’s capabilities. By grounding LLM responses in Knowable’s meticulously organized data, the new tools allow users to ask natural language questions and receive accurate, actionable answers.
Users can ask questions about specific documents.
As mentioned above, key features of the Ask Knowable suite include AI Contract Search, AI Contract Summaries, and AI Chat. Driving these features are several characteristics that help simplify search and enhance results:
- Natural language search. With Ask Knowable, users can pose plain-language questions, such as, “Give me the most recent active contract with Google,” or “What are the enforceable indemnification provisions in this contract family?” Unlike traditional search tools that require knowledge of Boolean operators or specific filters, Ask Knowable is able to interpret natural language queries and deliver results in seconds.
- Family mapping. Not to be redundant on the issue of family mapping, but Reed said that Knowable’s proprietary family-mapping algorithms are critical to the effectiveness of its AI. This structure enables users to see how terms evolve over time and how one document impacts others. For example, a user can determine whether an amendment affects a master agreement or only applies to specific SOWs. “Think of a family as the capital C contract, the parent agreement with all of its children,” Reed said.
- Document summaries. The system generates AI-powered summaries of individual agreements, highlighting key terms, parties, and provisions. This feature is particularly useful for quickly understanding complex documents without reading them in full.
- Multi-document analysis. Users can analyze entire contract families to answer broader questions. A user might, for example ask, “What are the current termination rights across all agreements in this family?” or, “Which agreements include force majeure clauses or similar terms?”
- Interactive responses. The tool not only provides answers but also links users directly to the relevant documents and clauses, allowing for deeper exploration.
So how can this be used? During the demonstration, Reed showed me several examples of how Ask Knowable can be used in real-world applications:
- Finding specific agreements. A user can quickly locate all active agreements with a specific counterparty, such as Microsoft, across a large repository.
- Analyzing term changes. The tool can show how contract terms, such as indemnification or payment provisions, have evolved across amendments.
- Force majeure and termination. Legal teams can identify all agreements with force majeure clauses or similar termination terms, a use case that gained prominence during the COVID-19 pandemic.
Closed Data Environment
Knowable is fully cognizant of the challenges associated with LLMs, particularly their potential for generating inaccurate or misleading responses, Reed said. “LLMs are designed for fluency, not accuracy.”
He said the company mitigates these risks in several ways:
- Closed data environment. The AI operates exclusively within the client’s contract repository, reducing the likelihood of irrelevant or incorrect results. “The large language model has a lot more context and knowledge than it would just if it was in the wild,” Reed said.
- Grounding in essential data. LLMs are anchored to the clean and accurate data Knowable has extracted from a company’s contracts, ensuring reliable outputs.
- Expert-driven prompt engineering. Legal experts and data scientists collaborate to design prompts that guide the AI’s reasoning.
Bottom Line
The Ask Knowable suite of AI tools will enter alpha testing with select clients after Thanksgiving. A broader beta release is planned for early 2025, with general availability expected around the time of the Legalweek conference in March 2025.
From what I saw during my demonstration with Reed, Ask Knowable represents a notable advance in the integration of AI into legal workflows, and specifically enterprise contract management.
Most notably, Knowable is taking a product that already used rigorous data management and traditional machine learning to manage complex contracts, and layering on generative AI in ways that appear to have real promise to make contracts more accessible and understandable.
