๐ก Inspiration
Negotiations in the aerospace industry are high-stakes, where a 1% difference in lead time or cost can impact multi-million dollar satellite projects. We realized that procurement professionals are often "flying blind"โjuggling static spreadsheets and historical KPIs while trying to maintain rapport in live video calls. We wanted to build a "co-pilot" that gives negotiators superhuman recall and real-time strategic insights without ever leaving the meeting.
๐ค What it does
NeGOATai is an AI-powered negotiation assistant tailored for aerospace procurement.
- Invisible Intelligence: Built on a custom fork of Free-Cluely, it exists as a translucent overlay on your desktop.
- Real-Time Context: Using Kontext.dev and Weaviate, it pulls live supplier data (quality scores, defect rates, and pricing trends) during negotiations.
- Human-in-the-Loop: It analyzes the live dialogue and suggests specific negotiation levers (e.g., "Ask for a 5% discount based on their 98% on-time delivery record").
- Video Integration: Leverages the Beyond Presence API to directly integrate into supplier video calls, creating a seamless interface between the negotiator, the supplier, and the AI.
๐ ๏ธ How we built it
- The Brain: We used OpenAI's GPT-4o for reasoning and dialogue generation.
- Knowledge Retrieval: We integrated Kontext.dev to give the AI contextual awareness, and Weaviate as our vector database to store and retrieve KPIs from a mock dataset of 10 aerospace suppliers.
- The Interface: The frontend was quickly prototyped using Lovable, providing an intuitive, professional-grade dashboard for procurement teams.
- The Framework: We customized Free-Cluely (Electron/TypeScript) to manage screen capture, OCR, and the "invisible" UI overlay.
๐ง Challenges we faced
Our biggest challenge was latency. In a live negotiation, an insight that arrives 10 seconds late is useless. We had to optimize our RAG pipeline through Kontext.dev to ensure supplier data was retrieved and processed fast enough to feel natural in conversation. Managing window focus and translucent overlays in Electron across different operating systems also required significant debugging.
๐ Accomplishments We're Proud Of
- Winning Best Usage of Kontext.dev: Successfully implementing real-time data retrieval that truly adds value in high-pressure scenarios.
- Aerospace Domain Focus: Creating a specialized mock dataset that reflects real-world aerospace KPIs like "Obsolescence Risk" and "AS9100 Compliance."
- Seamless Integration: Making the "Invisible Assistant" work smoothly with the Beyond Presence video interface.
๐ What we learned
We learned that the real power of AI in procurement isn't just "automation"โit's augmentation. By providing the right data at the right moment, we can turn a standard bargaining session into a strategic, data-driven conversation. We also gained a deeper understanding of building cross-platform desktop apps with Electron and the potential of low-code tools like Lovable for rapid UI development.
๐ What's next for NeGOATai
- Live Audio Transcription: Moving beyond screenshots to direct system-audio processing for faster responses.
- Multi-Supplier Comparison: Enabling the AI to pull "Battle Cards" that compare current suppliers against competitors in real-time.
- Tacto Integration: Strengthening the connection with Tactoโs existing procurement workflows to shift from "insights" to "automated PO updates."
Log in or sign up for Devpost to join the conversation.