Caylent Catalysts™
Generative AI Strategy
Accelerate your generative AI initiatives with ideation sessions for use case prioritization, foundation model selection, and an assessment of your data landscape and organizational readiness.
Venminder Unlocks 70% More Time for Risk & Compliance Analysis by Automating Data Retrieval with Generative AI
Artificial Intelligence & MLOps
Generative AI
At a glance
Venminder, an Ncontracts Company, streamlines document processing and compliance assessments by automating workflows using AI-powered solutions, unlocking 70% of their analysts’ time and reducing contract review turnaround from 65 days to under 3 days, while improving accuracy and scalability.
Solution Implemented
→ Augmented Venminder’s document processing workflow with an AI-powered solution using Amazon Bedrock and Anthropic Claude, automating data extraction and accelerating compliance assessments.
→Automated contract analysis with customized prompts and business rules for better accuracy in extracting key data points, such as start/end dates and renewal clauses.
→ Added OCR for scanned document processing to handle complex unstructured formats.
→ Used Amazon Kendra and S3 for document storage, indexing, and quick retrieval via natural language queries.
→ Enabled real-time processing, allowing parallel document analysis to eliminate backlogs and boost efficiency.
Outcomes Expected
→ 65-Day Backlog Cleared in 4 Days: Reduced the contract review backlog from 65 days to a 3 day backlog with real-time processing. This acceleration was achieved within 4 days of incorporating GenAI into their workflow.
→ 70% Time Saved for Analysts: Freed up 70% of analysts’ time by automating manual data retrieval.
→ More than 5X Faster Document Review: Reduced review time from hours to minutes, speeding up compliance assessments.
→ 86% Data Accuracy: Achieved 86%+ accuracy in contract data extraction, surpassing the 75% target.
A leap in efficiency, unlocked by generative AI
Venminder, an Ncontracts company, offers SaaS solutions that simplify third-party risk management, streamlining vendor onboarding, risk assessments, and compliance management. By automating key processes, Venminder helps organizations reduce the complexity of managing vendor relationships, ensuring stronger security and compliance.
Venminder’s success hinges on efficiently processing massive amounts of unstructured documents to ensure clients meet stringent compliance standards such as SOC 2. The company was looking for a way to automate the data retrieval process and the completion of compliance assessment reports with the processed data.
Caylent's generative AI powered solution introduced a faster, automated approach, transforming how Venminder conducts its assessments. Data Retrieval and answering compliance queries that previously took hours can now be accomplished in minutes. It allows Venminder to scale without increasing staff and improves document review times by over 5X, significantly boosting operational efficiency. Venminder’s clients benefit from quicker, more accurate compliance assessments as well as access to new insights, helping them meet critical regulatory deadlines.
In addition to a document processing and data retrieval solution, Caylent also helped Venminder automate contract analysis by modernizing an internally developed application. By migrating from OpenAI to Anthropic Claude on Amazon Bedrock and with sophisticated prompt engineering and business rule configurations, the system is able to achieve 86%+ accuracy in extracting data points such as start/end dates, automatic renewal terms, and clauses. This system allowed Venminder’s analysts to quickly review contracts and flag important data points, leading to faster, more efficient contract management.
Problem
Venminder’s core challenge was twofold.
1. Document Processing for Data Extraction
Venminder’s analysts were manually processing vast volumes of unstructured documents for third-party risk management reviews. This labor-intensive process resulted in bottlenecks, high turnaround times, and increased risk of human error. With each document requiring thorough review, scalability became a growing concern as their client base expanded. Analysts were spending 70% of their time on data retrieval instead of more meaningful risk assessments, causing inefficiencies.
This was especially problematic during certain periods like the start of the year when Venminder typically experiences a spike in contracts. Venminder was facing troubles keeping up with client demand without a way to accelerate data retrieval.
2. Inconsistent Contract Analysis
Venminder's initial implementation of OpenAI for contract analysis failed to meet their accuracy expectations, achieving only 60% precision in extracting essential contractual data. It failed to capture essential data like complex data formats and clauses.
This low accuracy led to prolonged manual validation, preventing the company from leveraging the full potential of AI automation. The company’s contract backlog had escalated to over 65 weeks during high traffic periods. Venminder needed a solution to both improve accuracy and support complex contract analysis at scale.
Solution
Automated Document Processing
Caylent introduced a document processing solution leveraging Amazon Kendra. This automated system allowed Venminder to ingest, index, and search large volumes of unstructured data from various formats. The use of AWS Bedrock enhanced search functionality by allowing natural language queries, significantly improving data retrieval efficiency.
Caylent incorporated Optical Character Recognition capabilities for scanned documents more effectively. The system was also fine-tuned to improve accuracy and manage the complexities of dealing with many documents at once.
Key solution components:
- Amazon S3 & Kendra for document storage and indexing: Documents were stored in S3 buckets and indexed by Kendra, allowing for near-instant retrieval through simple language queries.
- AI-powered search using AWS Bedrock: Integrated with Kendra, AWS Bedrock powered natural language queries, enabling analysts to interact with the system as if they were speaking to a colleague, vastly reducing the complexity of the search process.
The solution reduced document review time by more than 70%, allowing Venminder’s analysts to focus on critical data points rather than spending hours manually combing through documents.
Enhanced Contract Analysis
Recognizing the limitations of the OpenAI-powered solution, Caylent transitioned Venminder’s contract analysis to Amazon Bedrock. The new architecture improved the system’s ability to extract key information, such as contract start/end dates, renewal terms, and critical obligations.
Key solution components:
- Tailored prompt engineering: Caylent customized the prompts for AWS Bedrock to meet Venminder’s specific requirements, improving accuracy from 60% to 86%+.
- Custom logic and business rule integration: The system was configured to deliver results in Venminder’s desired format, including standardized date formats and clear yes/no answers for renewal clauses.
- Real-time processing: Documents were analyzed in real-time, cutting down analysis time from hours to minutes.
By transitioning to AWS Bedrock, Venminder achieved far more consistent results, reducing manual oversight and enabling their analysts to process contracts faster and with greater accuracy.
![Perform[cb] Logo](https://cdn.builder.io/api/v1/image/assets%2F78dfb6aee9fd4aa4be447bb5ccea13e3%2F3bd5083b2bdd403b940a5ac56631df5f?width=640&quality=90&format=webp)
