DocuPilot
From documents to decisions.
Status: Team prototype · public demo · first place at the AI Innovation Bootcamp
An award-winning team prototype that turns client requests, contracts and invoices into requirements, risks and decisions. Abdullah Misar led the team and contributed AI generation, shared data APIs and responsive workflows.
The problem
Project decisions get scattered across client messages, requirements, contracts and invoices. Teams need a connected view of what was agreed, what changed and what needs attention.
My role
Team lead & full-stack contributor · AI generation, data APIs and product workflows
What I worked on
- Gemini-powered SRS generation and conversational refinement from Arabic or English client requests, with schema validation and optional Supabase persistence.
- Project Intelligence and Scope Guard integrations that turn requests into structured project assessments, scope changes and recommended actions.
- Shared domain types, normalized AI outputs and project read APIs connecting documents, risks and approvals in the demo workspace.
- PDF, TXT and Markdown extraction, plus document-upload integration and graceful handling of missing retrieval context in Ask DocuPilot.
- Responsive interfaces, working local approval and invoice actions, downloadable dashboard and risk CSVs, and interaction tests with seeded demo data.
Built with a team that won first place at the AI Innovation Bootcamp by GDG on Campus University of Jeddah and StartSmart. This is a demo-ready prototype: the public dashboard uses sample data, several workspace actions persist locally, and production authentication and multi-tenant workflows remain unfinished. The retrieval engine and provider-reliability work include other team members’ contributions.


