Inspiration
Companies spend countless hours on repetitive tasks such as checking purchase orders in customer portals and manually entering that information into their own systems. These processes are time-consuming, prone to human error, and difficult to scale. We were inspired by the idea of creating an AI-powered solution that not only automates tasks but also understands the context behind them, allowing it to adapt to changes in workflows and interfaces.
What it does
ShelfIQ is an intelligent automation platform that learns business processes from a single manual demonstration. Once a user performs a task, the AI analyzes the workflow, understands the context of the information being handled, and can autonomously execute the same process in the future.
In our use case, ShelfIQ enables suppliers such as Arca Continental to automatically retrieve purchase orders from customer portals and transfer them into internal systems, eliminating repetitive manual data entry and reducing operational workload.
How we built it
We built ShelfIQ using:
React for the user interface. FastAPI for the backend and process orchestration. AI models for contextual understanding and workflow interpretation. Web automation technologies to interact with external systems. Intelligent data extraction to identify and process purchase order information.
The platform records an initial workflow demonstration and leverages AI to replicate and adapt the process in future executions.
Challenges we ran into
Building a solution that does not rely on fixed UI elements or page layouts. Enabling the AI to identify relevant information even when websites change. Combining web automation with contextual AI reasoning. Ensuring reliable data extraction and transfer between different systems. Developing a learning mechanism that can generalize from a single demonstration.
Accomplishments that we're proud of
Creating a system capable of learning a business workflow from just one demonstration. Reducing manual effort for repetitive operational tasks. Implementing context-aware automation instead of traditional rule-based automation. Delivering a functional prototype within a limited timeframe. Demonstrating how AI can make enterprise processes more efficient and scalable.
What we learned
Traditional automation is highly dependent on static interfaces and can easily break when systems change. Context-aware AI creates more resilient and adaptable automation solutions. User experience is critical when designing workflow training and automation tools. Real-world business challenges require flexible systems that can understand intent, not just predefined rules. AI agents can significantly improve operational efficiency when paired with practical business workflows
What's next for ShelfIQ
Support multiple variations of the same workflow. Continuously improve performance through feedback and learning. Integrate with additional ERP, CRM, and enterprise platforms. Add monitoring, auditing, and analytics capabilities for business users. Evolve into a fully autonomous AI agent capable of managing end-to-end procurement and order management processes with minimal human intervention.
Log in or sign up for Devpost to join the conversation.