The entire contract process requires a great deal of precision through its various stages–initiation, authoring, workflow, negotiation and approval, execution, compliance, and contract renewal. To avoid human errors and increase efficiency and productivity, several organisations are opting for automated and streamlined Contract Lifecycle Management (CLM) systems. Such systems create a common language, standardise the contract development process, and reduce the time taken by using pre-approved templates and legal clauses.
SirionLabs, a SaaS contract lifecycle management (CLM) platform, was named as a leader in CLM solution providers in the Forrester Wave 2021 report. Analytics India Magazine caught up with Aditya Gupta, Chief Architect and Co-founder of SirionLabs, to understand its services and how AI plays a major role in their functioning.
Sign up for your weekly dose of what's up in emerging technology.
AIM: Tell us about the main services and offerings of SirionLabs
Aditya Gupta: SirionLabs is a leading SaaS contract lifecycle management (CLM) platform that uses AI to help businesses manage the complete contracting lifecycle on a single, easy-to-use platform.
Our platform’s AI core is one of the most advanced in the industry, and we use it to help organisations digitally transform traditional CLM functions such as contract negotiation, authoring, repository, and extraction of contractual data. In addition, our platform also offers post-signature contract management capabilities focused on supplier governance, obligation and service level management, automated invoice validation, and data-driven buyer-supplier collaboration.
We have made significant investments towards developing AI’s role in our contract management systems further. Sirion has helped clients unlock significant enterprise value including hard savings of 10-12 percent of the contract value, improved business outcomes, up to 80 percent reduction in customer-supplier disputes, and a 60 percent reduction in manual effort and cost associated with contract creation and governance.
AIM: What are the proprietary algorithms and AI/ML models currently used At SirionLabs?
Aditya Gupta: Sirion’s core is built on a foundation that fuses machine learning (ML) and natural language processing (NLP). We use supervised learning coupled with a massive corpus of industry data to train our ML models to recognise and ‘read’ not just paper contracts and PDFs but also handwritten notes. Our image recognition and OCR technologies together play a vital role in digitizing contract documents, which are then semantically parsed by our NLP engine to break them down into machine-readable text.
Post digitisation, our platform uses an AI-based expert system to solve complex decision-making problems such as helping users in identifying relevant clauses. Our platform also uses its NLP and ML capabilities to accelerate the legal review process for counterparty papers by automatically ‘reading’ contracts and flagging deviated and missing clauses, and process natural language queries from users, based on which it provides analytical insights and data visualisations.
AIM: What comprises the tech stack at SirionLabs?
Aditya Gupta: We widely use spaCy, Pytorch, and TensorFlow libraries to support our AI and ML initiatives. Besides these, we use various other libraries for logic building and transformation.
Speaking of frameworks, Sirion uses a shared ledger whose architecture is similar to those used in blockchain applications, and we use it to manage the flow of engagement data between counterparties. This framework has helped us position Sirion CLM as a frontend collaboration platform that acts as a single source of truth for performance data for all parties in a contract. Our goal is to help customers end their dependence on traditional siloed backend systems – like ERP and ITSM – that usually host all sorts of critical supplier records.
Sirion is a multi-tenant SaaS-based application, built for scale and security from the ground up. It is a modern MVC architecture with some business microservices. It is based on Java, Spring, PostgreSQL and Elasticsearch. While Java is our primary programming language, we also use .Net, Golang and Python.
AIM: Tell us about SirionLabs’ M&A solution
Aditya Gupta: Merger and acquisitions are becoming an increasingly common way for enterprises to grow and acquire new lines of business or products. However, they are also among the most complex events an organisation can go through. Entire business systems need to be integrated and hundreds of due diligence queries need to be answered. Most companies usually depend on labour and time-intensive manual review processes to identify key risk elements in contracts.
Sirion’s M&A solution combines AI-led contract extraction, analytics, and authoring capabilities to automate large swathes of due diligence, risk mitigation, and repapering activities that underpin a typical M&A event.
During an M&A, the diligence team can upload contracts and associated documents in bulk and use Sirion’s out-of-the-box adaptors to connect with other enterprises’ IT systems and ingest documents stored in these data silos. Our platform’s AI engine then converts all legacy contracts, PDF files, and even handwritten notes into machine-readable text, which is then semantically parsed to extract key metadata, service levels, obligations, pricing tables, and more. At this stage, a similarity clustering algorithm helps users identify and remove duplicate documents from the entire corpus. The extracted data can then be tagged and batched into logical groups and assigned to human reviewers, who can then normalize the results.
After the basic review process, Sirion’s AI can scan through the digitized contracts to highlight missing clauses and deviations, based on which contract managers can choose to repaper or novate inherited contracts to bring them in line with the post-M&A state.
AIM: What skills you look for in candidates applying for AI and Data Science-related roles?
Aditya Gupta: Currently, we are working on expanding our workforce in the US to drive key AI initiatives out of our new CoE in Seattle. This has given us a clear view of the technical and individual qualities that we would like to have in our engineers.
We look for machine learning research engineers who have practical experience in NLP, streaming applications (Kafka, Rabbitmq), programming languages such as Python and R, and deep understanding of statistics, data structures and algorithms, linear algebra, differential calculus, optimization and numerical analysis, advanced mathematical modelling, and comparative KPI analysis.
While there is significant overlap between an AI engineer’s and data scientist’s respective skill sets, the latter’s role requires experience in data mining; R, SQL and Python; familiarity with Scala and Java; and extensive working knowledge of business intelligence tools such as Tableau and data frameworks such as Hadoop.
We have an inclination towards people with a growth mindset who thrive in a collaborative environment that focuses on solving real-world business problems.
AIM: What are your company’s short and long term goals on the tech front?
Aditya Gupta: Our current goal is to democratize CLM, which we hope to achieve by launching:
- Sirion’s AI engine as a standalone self-service solution
- A developer platform that will include public APIs, SDKs
- An app marketplace for first (created in house) and third-party apps (created by a developer community).