Kuna | Devpost

Inspiration

Small wholesalers face uncertainties both from the suppliers they connect with and the many diverse customers they serve. Customers often communicate in scattered messages across WhatsApp, email, phone, and other channels for placing orders. This creates error-prone manual work. Orders are often lost or miscommunicated due to these patterns, leading to gaps in supply or wasted items due to over-ordering. This presents an opportunity for AI agents to orchestrate every interaction with customers, building an advanced profile of the customer that allows them to connect and re-order with ease.

What it does

Customers can WhatsApp message their order to our wholesaler even in the form of an image. A dedicated AI agent takes care of them, making sure their order fills their needs and fits their unique communication style. Wholesalers can easily track relationships with customers, monitoring sales growth, recent communications by agents, and trends in relationship health. Anomalies in order habits are flagged in order to ensure we deliver the goods to the customers they really want that fit their usage profile.

How we built it

Python, Next.js, WhatsApp Meta API, ngrok

Challenges we ran into

MetaAPI restrictions for non business accounts prevented easily texting from any number.

Accomplishments that we're proud of

Ability to analyse multimodal messages and generate orders from images. Texting via WhatsApp from multiple phones at once.

What's next for Kuna

We're ready to sell to customers, starting in our home city Oxford! Also, we’ll expand into the closest adjacent sectors where the same multi-channel ordering chaos and recurring B2B reorders make our agent immediately valuable (e.g., Building Materials & Construction Supplies Distribution, Packaging & Disposables Distribution). Our long-term visions is a horizontal agentic procurement “operating system” for Europe’s mid-size distributors, not just a Food & Beverages tool.