CoLabs, a software incubator, decided to develop IntelAgree—an application to help enterprises automate and optimize the contract management process. The vision for IntelAgree was to not only offer standard features like approvals and workflows, but also interpret contracts based on their key attributes and clauses. Realizing a traditional software approach was impractical, the IntelAgree team needed innovative strategies and tools to manage the high variability of unstructured text data within agreements. To do this, IntelAgree sought AgileThought’s data science expertise to develop a custom AI strategy and deploy a production-quality machine learning algorithm.
- Navigating artificial intelligence technologies and identifying a cost-effective, secure and scalable solution
- Finding and creating sources of training data for the machine-learning algorithm
- Defining an efficient, de-biased approach to accurately label unstructured text data in contracts
- Evaluating, selecting and deploying the appropriate machine-learning algorithm to improve the accuracy of extracted attributes and clauses within contracts
Paving the Path to Predictive Analytics
AgileThought conducted Predictive Analytics Discovery—a proprietary methodology for analyzing data, statistically validating data quality and determining machine learning readiness—and evaluated a variety of data science methods, tools and techniques to define IntelAgree’s custom AI strategy. AgileThought used tools inspired by cutting-edge research on natural language processing, which not only gave IntelAgree a performance advantage, but also taught them how to apply research to product development.
Implementing a Cutting-Edge AI Strategy
Using a large, open-source dataset of legal contracts, AgileThought partnered with IntelAgree to manually label and validate key attributes (i.e. effective date, contracting parties) and clauses (i.e. confidentiality, severability) to train the machine learning algorithm. Using a combination of open-source components and Microsoft technologies such as Azure Machine Learning Service, Azure VMs and Azure Blob Storage—AgileThought partnered with IntelAgree’s .NET and web engineering team to build the AI solution. Using a combination of machine learning and natural language processing techniques, the platform now automatically interprets uploaded contracts and extracts essential information
Reducing Contract Processes from Hours to Minutes
Now, IntelAgree can instantly identify attributes across a variety of contracts—reducing processes that previously took 4-5 hours down to 5-30 minutes. With the platform’s searchable repositories and machine learning capabilities, users can query terms like, “contains a confidentiality clause,” to refine results faster. The platform also offers dashboards for clients to track key metrics like, expiration dates of “in force” contracts, or comparisons of payment terms across different contracts. Because of this, client attorneys can now spend more one-on-one advisory time with clients and less time reviewing the minutiae of contracts.
David Hull, CEO of CoLabs