Privacy-First AI Legal Tech Case Study: Ookulli on AWS

Traditional legal reviews are often too expensive for freelancers, forcing many to sign risky contracts blindly. To solve this, Ookulli built a secure, event-driven architecture on AWS that anonymizes sensitive data before utilizing Amazon Bedrock for structured, clause-by-clause analysis.

ClientOokulli

PublishedJuly 2026

IndustryLegal

Websitehttps://www.ookulli.com/

Discover how a rising legal tech platform leverages advanced AI to provide instant, budget-friendly contract reviews without sacrificing user privacy.

Ookulli is a UK-focused, privacy-first contract review platform built for freelancers and individuals who are asked to sign contracts, most often employment agreements, without the time, budget, or legal background to properly assess the risk. The product originally targeted rental agreements, but the team moved into freelance and employment contracts after finding that renters, constrained by the housing market, had little real negotiating power and would often sign regardless of risk. Freelancers and employees, by contrast, have more room to review, question, and negotiate before signing, which made the product's value far easier to prove.

Freelancers and individuals receiving contracts have historically faced three unappealing options.

Cost Barrier: Traditional legal review can cost hundreds of pounds, pricing many freelancers out of getting a contract checked at all.
Blind Signing: Without an affordable alternative, many people sign contracts without properly understanding the risk they are taking on.
Unsafe Generic AI: Generic AI tools fill some of the gap, but they are not private enough for sensitive contract data, and they can hallucinate or focus on the wrong clauses entirely.

Behind the simple user experience lies a modern event-driven architecture running on AWS.

Documents uploaded by customers trigger an automated processing *workflow built on Amazon ECS, AWS Fargate, Amazon EventBridge, Amazon DynamoDB, and Amazon S3. *
Since AI-based document analysis can take longer than the execution limits of traditional serverless functions, the platform relies on containerized workloads to process requests reliably at scale.

As AI usage continued to grow, Ookulli wanted to validate that both the application architecture and its Amazon Bedrock integration would remain efficient and cost-effective. They also wanted to improve observability, strengthen the security of sensitive legal documents, and prepare the platform for future customer growth.

The customer created a solution on AWS for law document processing with these steps:

Privacy-First AI Processing: Documents are anonymised using AWS services (Amazon ECS and Amazon SageMaker) before analysis and processed securely within a European AWS Region.
Structured Legal Analysis: Amazon Bedrock generates a structured review highlighting key risks, negotiation points, and clause-level insights.
Transparent Results: Every finding is linked to the relevant contract section, allowing users to ask follow-up questions before signing or consulting a lawyer.

Stormit conducted a comprehensive AWS Well-Architected Review, focusing on the platform's AI infrastructure, container architecture, and future scalability.

Together with the customer, we reviewed how Bedrock could support contract summarization, clause detection, and natural-language explanations while providing a managed and scalable foundation for future AI development.
We also discussed the evaluation of additional foundation models for more advanced legal reasoning, such as detecting contradictions and inconsistencies across complex contracts.
Recommendations included validating the Amazon ECS and AWS Fargate architecture, improving monitoring with Amazon CloudWatch, strengthening encryption of customer documents using AWS KMS, and identifying opportunities for future cost optimization through AWS Savings Plans and efficient Amazon Bedrock usage.

By validating the architecture before rapid growth, the team gained confidence that the platform could efficiently support increasing AI workloads while following AWS best practices for security, reliability, and operational excellence.

Reviewed 10+ AWS services as part of the Well-Architected Review.
Optimized an AI platform powered by Amazon Bedrock and Anthropic Claude.
Identified 15+ architectural recommendations across security, reliability, operational excellence, and cost optimization.
Validated a scalable event-driven architecture built on Amazon ECS and AWS Fargate.
Prepared the platform for future AI workload growth before public launch.

The review also established a strong foundation for expanding the use of Amazon Bedrock and Anthropic Claude models as the platform evolves, helping Ookulli continue delivering fast and intelligent legal document analysis to its customers.

AWS SERVICES

Amazon Bedrock

Amazon ECS

AWS Fargate

Amazon SageMaker

AWS KMS

Key Takeaways

10+

Services Reviewed

15+ Recommendations

Identified

Minutes

Delivery

Igor is our go-to person for cloud services.

Ask him anything.