## Inspiration
Enterprise software procurement is often fragmented and time-consuming. Procurement teams spend hours researching vendors, comparing pricing, verifying compliance, assessing risks, and coordinating approvals across multiple platforms. These repetitive tasks slow down purchasing decisions and make it difficult to make informed, evidence-backed choices.
We wanted to build a solution that streamlines this entire workflow. Instead of relying on scattered research and manual evaluation, we envisioned an AI-powered procurement operating system that acts as a team of specialized procurement analysts, helping organizations make faster, smarter, and more transparent software purchasing decisions.
## What it does
ProcurementOS is an AI-powered procurement platform that automates the software vendor evaluation process.
Users begin by submitting a procurement request, after which multiple AI agents work together to:
- Research software vendors using real-time web data
- Evaluate security, compliance, and business risks
- Compare pricing, features, and vendor capabilities
- Generate evidence-backed recommendations
- Score vendors based on ROI and procurement criteria
- Trigger procurement workflows through Workato integration
- Monitor vendors for future risks and changes
The platform presents all findings through an interactive dashboard, enabling procurement teams to make confident decisions within minutes instead of days.
## How we built it
We developed ProcurementOS using Python and Streamlit as the primary application framework.
Our solution combines several technologies:
- Exa API for real-time vendor research and evidence gathering
- Workato for workflow automation and procurement execution
- Plotly for interactive analytics and visualisations
- Python to coordinate multiple specialized AI agents responsible for research, risk analysis, ROI scoring, monitoring, and recommendations
Rather than relying on a single chatbot, we designed a multi-agent workflow where each agent focuses on a specific procurement responsibility before combining their outputs into one recommendation.
## Challenges we ran into
One of our biggest challenges was designing a workflow that felt like a real procurement process instead of simply wrapping an LLM around a prompt.
We also had to balance live API responses with a smooth user experience. External services can occasionally be slow or unavailable, so we implemented fallback mechanisms to ensure the application remained functional during demonstrations.
Another challenge was presenting large amounts of procurement data in a way that was easy to understand, which required multiple iterations of our dashboard and scoring system.
## Accomplishments that we're proud of
We're proud of building an end-to-end procurement platform rather than a single AI feature.
Some highlights include:
- Designing a multi-agent procurement workflow
- Integrating real-time vendor research with evidence-backed recommendations
- Automating procurement execution through Workato
- Building an interactive dashboard for vendor comparison and monitoring
- Creating a polished prototype that demonstrates how AI can augment enterprise procurement instead of replacing human decision-making
## What we learned
This project taught us that building useful AI applications is less about prompting a language model and more about designing reliable workflows around it.
We also gained hands-on experience integrating multiple APIs, handling external service failures gracefully, designing intuitive user interfaces, and orchestrating specialized AI agents into a cohesive system.
Most importantly, we learned how to build AI systems that support human decision-making through transparency and evidence rather than simply generating answers.
## What's next for SixSeven
We envision SixSeven evolving into a comprehensive procurement platform for enterprises.
Future plans include:
- Integration with ERP systems such as SAP and Oracle
- Support for contract analysis and negotiation assistance
- Custom procurement policies tailored to different organisations
- Continuous vendor monitoring with automated alerts
- Predictive budgeting and spend forecasting
- Collaboration tools for procurement teams and stakeholders
- Deployment-ready authentication and role-based access control
Our long-term goal is to transform software procurement from a manual, fragmented process into an intelligent, transparent, and data-driven workflow powered by AI.
Log in or sign up for Devpost to join the conversation.