Adjacent
Tagline: Turn Any Meeting or Document into Actionable Work
Adjacent helps teams automatically extract tasks, assign responsibilities, and organize work from Zoom or Google Meet meetings, PDFs, and PowerPoint presentations. Say goodbye to messy notes and forgotten follow-ups.
Problem
There is a massive amount of information missing outside of traditional Product Management tools. SO by turning conversations and content into actionable tasks, important decisions, dependencies, and ownership don't slip through the cracks — leading to faster delivery and increased productivity.
Our Solution
Adjacent automatically turns multi-format inputs (meetings, slides, PDFs) into a prioritized, assigned Kanban board using agentic AI. Upload or connect content, let the system transcribe and analyze it, and get a ready-to-use task board with suggested assignees and deadlines — all in minutes, not hours.
How it works
- Ingest — Upload recordings, transcripts, PDFs, or slide decks (or connect Zoom/Google Meet).
- Transcribe — Whisper converts audio/video into clean transcripts.
- Analyze — Nemotron agents parse text and documents to extract tasks, dependencies, and risks.
- Assign — Tasks are matched to people using historical workload and availability signals.
- Present — Tasks appear in an editable Kanban view with priorities, deadlines, and chat-driven edits.
Tech stack
- Frontend: Next.js 14, React, TypeScript, Tailwind CSS
- Backend: Flask API for media processing and agent orchestration
- Database & storage: Supabase (Postgres, Auth, Storage)
- AI: Nemotron agents for task extraction; Whisper for transcription
- Integrations: Zoom and Google Meet APIs
Key benefits
- Time saved — turns hours of manual work into minutes.
- Better decisions — captures actions, owners, and dependencies automatically.
- Agile-ready — instantly usable Kanban-style boards with smart priorities.
- Multi-format support — works across meetings, PDFs, slides, and code repos.
Use cases
- Daily stand-ups, sprint planning, retrospective actions
- Cross-functional project reviews and handoffs
- Remote teams needing reliable meeting-to-action capture
Accomplishments
- Fully functioning prototype handling multi-format inputs and AI task extraction
- Nemotron agents autonomously proposing and assigning tasks
- Real-time task tracking and metadata stored in Supabase
What We Learned
-Integrated Supabase for smooth database management and real-time data syncing. -Successfully connected and deployed NVIDIA Nemotron for intelligent task extraction and analysis.
What’s Next
-Improve AI accuracy for more precise task and context understanding. -Redesign the user interface for a cleaner, faster experience. -Optimize backend performance to make task generation and display faster. -Add an interactive chat bot feature for real-time edits and team collaboration.
Team
Jaiom, Saptarishi, Matthews, Jordan
Built With
- nvidia-nemotron
- python
- react-js
- supabase-db



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