WarehouseWise — inventory counting on Bee

Inspiration

We noticed that manual inventory counting is prone to error and incredibly time-consuming. We wanted to show how the Bee track could be used to solve industrial logistics challenges using intelligent agents. Our goal was to create a tool that makes warehouse management accessible to developers.

What it does

WarehouseWise performs automated inventory counting by processing warehouse data and visual inputs. It uses AI to identify mismatches between physical counts and digital records. The platform allows developers to integrate these insights directly into existing management workflows.

How we built it

We implemented the logic using TypeScript and utilized AWS Lambda for serverless execution. We used Langgraph to manage the stateful logic of our counting agents and Amazon Bedrock for natural language processing. The entire solution is hosted on the Bee surface to ensure seamless integration for developers in this track.

Challenges we ran into

Managing the state across complex inventory cycles in Langgraph was initially difficult to map. We also struggled with ensuring the Amazon Bedrock models correctly interpreted various warehouse data formats. Refining the prompts to ensure high accuracy in inventory counting required several iterative tests.

Accomplishments that we're proud of

We are proud of how seamlessly we integrated multi-agent workflows within the Bee environment. Seeing the system autonomously resolve complex inventory discrepancies was a major milestone for us. Our TypeScript implementation is modular and easy for other developers to extend.

What we learned

We learned a lot about orchestrating sophisticated AI agents using the Langgraph framework. We also gained deep insights into how the Bee track can empower enterprise-grade applications. Understanding the nuances of AWS Lambda cold starts helped us optimize our architecture significantly.

What's next for WarehouseWise

We plan to add real-time video feed processing for even faster inventory counting. We also want to expand the analytics capabilities to predict future stock needs. Our ultimate goal is to make WarehouseWise the standard toolkit for developers building on Bee.

Built With

• amazon-bedrock
• amazon-bee
• aws-lambda
• inventory
• langgraph
• warehousewise