Building LegalLens.ai involved multiple technical challenges. One major challenge was ensuring consistent and reliable clause-level analysis from the AI, as large language models can sometimes produce unstructured or inconsistent outputs. We addressed this by enforcing a strict output format and mapping the AI responses into predefined legal categories and risk levels.
Another challenge was balancing accuracy and explainability. Instead of simply flagging clauses as risky, we focused on making the reasoning understandable to non-legal users. This was solved by designing prompts that require the AI to highlight trigger terms and explain risks in plain language. Managing time constraints while integrating AI logic, UI, and demo-ready workflows within a short hackathon window was also a challenge, which we overcame through clear task division and iterative testing.