What Termly Does
- Core Value: An AI-powered contract and vendor lifecycle management platform. Users drag-and-drop a contract or invoice PDF, and in under 60 seconds, our AI pipeline (Textract + Bedrock) extracts key clauses, obligations, and dates. It maps upcoming liabilities on a clean dashboard, schedules renewal alerts, and generates automated vendor risk scores.
Inspiration
- The Spark: Rooted in firsthand experience watching family and friends run businesses where vendor contracts, obligations, and renewals are handled manually (scattered in printouts, email threads, and spreadsheets). The direct cost of forgotten renewal windows and payment tracking is a real loss we wanted to solve for SMBs.
How We Built It
- Frontend Speed: Designed and bootstrapped the frontend using v0, which made building the landing page and authentication UI a breeze.
- Serverless Backend: Built a highly scalable backend orchestrating AWS services (Lambda, API Gateway, S3, Aurora Serverless v2, Cognito) to ensure rapid response times and robust handling of document OCR (Textract) and AI analysis (Bedrock/Claude).
Challenges Faced
- Performance vs. Aesthetics: Marrying complex UI features (smooth transitions, dual-mode theme, micro-interactions, and Liquid Glass surfaces) with strict performance budgets (<2s dashboard loading times).
- System Integration: Connecting and securing the end-to-end serverless pipeline, ensuring reliable DB transactions, and isolating client data at the organization level (
org_idisolation) across all lambda operations.
Accomplishments We Are Proud Of
- Accessible & User-Friendly: Building a solution that is simple and friendly enough for small/medium business owners. Unlike enterprise software (which requires weeks of training), a non-technical manager can get first value from Termly in under 5 minutes.
- B2B Power in a Clean Interface: Successfully balancing advanced capabilities (AI clause extraction, automatic obligation mapping, and spend analytics) inside a clean, high-performance dashboard that does not feel cluttered.
What We Learned
- AI extraction pipelines require strict boundary constraints and fallback handling to maintain stability on poor document scans.
- Developing a clean, minimalist design language (Apple-style UX) requires more restraint and planning than over-designed layouts.
What's Next for Termly
- Custom Approval Workflows: Implement a multi-approver chain (up to 5 roles) to sign off on contracts before they shift from pending to active.
- SMS Alert Integrations: Extend SNS support to notify users of critical 7-day deadlines via text message.
- Deep Multi-Language Parsing: Training the Bedrock extraction prompt to support multi-lingual contracts and regional compliance frameworks.
Built With
- amazon-ec2
- amazon-web-services
- aws-aurora-postgres
- aws-bedrock
- aws-textract
- next.js
- typescript
- vercel


Log in or sign up for Devpost to join the conversation.