UniSphere: Transforming Public Sector Procurement with AI UniSphere is an AI solution revolutionizing government procurement for connectivity projects through automated RFP analysis and bid evaluation. The Problem Government procurement suffers from inefficient, error-prone processes that lead to delays, wasted funds, and poor vendor selection. Creating comprehensive RFPs and objectively evaluating bids remains challenging for procurement teams. Our Solution UniSphere's procurement-specific Language Intelligence Model (LIM) processes complex documents with precision, built on open-source technologies for transparency and flexibility. Key Features RFP Analysis: Automatic requirements and criteria extraction. Technical specification identification. Gap detection in vendor proposals. Bid Evaluation: Objective bid scoring against requirements. Detailed strengths/weaknesses analysis. Best practices integration. Technology Built using Llama 3.1, IBM Granite, Hugging Face, Docker, PyTorch, and FastAPI—ensuring security, scalability, and seamless integration. Benefits for Users. Procurement Officers: Faster, more efficient processes. Technical Evaluators: Consistent, objective evaluations. Security Officers: Secure, compliant implementations. Challenges and Roadmap Current Focus: Security through secure self-hosting. Developing robust procurement datasets. Creating continuous improvement mechanisms. Future Plans: Enhanced security features. Risk prediction and sentiment analysis. Human-in-the-loop accountability. Ongoing model refinement. Conclusion UniSphere transforms public procurement by automating critical processes, helping governments save time, reduce costs, and improve decision-making. Our open-source approach ensures transparency and adaptability, building a foundation for more efficient, accountable procurement practices in connectivity projects.
Category tags:"Very clear presentation and value proposition. The Demo was the same as the video that was uploaded. It would have been great to see a live prototype."
Jenny Tillgren
Venture Scout and Advisor
"Great product but needs demo"
Surabhi Nayak
Privacy Engineer
"Lots of great features, would be great to see how the fine tuned LLM performs against a non fine tuned model"
Thomas Blake
CTO