Quikka

Inspiration

Peer-to-peer social commerce is the backbone of the African retail economy, but it is bottlenecked by a massive trust deficit. The spark for Quikka came from personal experience: while in university, I was nearly scammed out of 12,000 shillings trying to buy a new pair of sneakers from an online vendor. On platforms like WhatsApp and Instagram, sellers and buyers face a daily dilemma. Buyers refuse to pay before delivery out of fear of being scammed, and sellers refuse to dispatch goods without upfront payment. The result is massive friction, canceled orders, and lost revenue. We realized these informal merchants didn't need a complex e-commerce storefront; they needed a simple, trustless financial primitive: escrow. We were inspired to build a system that meets these merchants exactly where they already work—on WhatsApp—while abstracting away the complex database infrastructure needed to secure their money.

What it does

Quikka is a distributed, AI-enhanced escrow ledger built for the informal African gig and social commerce economy. Operating natively via a WhatsApp bot, it allows sellers to instantly generate a secure checkout gateway for their items. Currently, merchants can simply text the bot to describe a product, and the AI automatically generates product links and appends the items to their web store.

When a buyer pays via local mobile money (M-Pesa), the funds are securely locked in Quikka’s AWS ledger, and the buyer receives a secure cryptographic PIN. The seller is instantly notified and dispatches the item. Upon delivery, the buyer hands the PIN to the seller, who inputs it into the WhatsApp bot to instantly release the funds to their wallet. It mathematically eliminates the "pay-on-delivery" flake rate and completely protects the buyer from fraud.

How we built it

To build a financial ledger, zero-downtime and absolute data integrity were non-negotiable. We leveraged the Vercel and AWS H0 stack to achieve this with zero operational overhead.

Our frontend and API routes are built with Next.js and deployed on Vercel, giving us instant global edge delivery and seamless CI/CD. For the core escrow state machine, we utilized AWS Aurora DSQL. Because we are handling real financial states (Locked, Released, Refunded), DSQL provided the serializable, distributed, and highly available transactional database we needed to ensure funds were never double-spent. We integrated the WhatsApp API for the conversational interface and IntaSend for localized M-Pesa payment processing.

Challenges we ran into

Building a financial application on a weekend timeline is inherently risky, and managing asynchronous state between a local telecom payment gateway (M-Pesa) and our serverless infrastructure was brutal. We had to squash a critical bug where the release PIN occasionally failed to resolve the escrow transaction.

Our biggest technical hurdle was handling network instability and concurrency. If the server temporarily went down before a transaction resolved, or if a buyer lost internet connectivity and reconnected, transactions would either fail or process twice. Fixing this required us to implement strict idempotency keys and rely heavily on Aurora DSQL's transactional guarantees to ensure the database state perfectly matched the real-world flow of money.

Accomplishments that we're proud of

We are incredibly proud of bridging enterprise-grade AWS infrastructure with the gritty reality of African street commerce. We successfully architected a system where a vendor holding a pair of sneakers is interacting with a highly scalable, serializable AWS database through a simple WhatsApp text. During our testing and initial rollout phase, we have successfully processed 15 live transactions, proving that with Vercel and AWS, a two-person team can build robust financial infrastructure that normally takes venture-backed startups months to deploy.

What we learned

We learned that in fintech, the architecture is the product. You cannot fake an escrow ledger. We also learned the power of "Framework Defined Infrastructure" through Vercel—being able to push Next.js code and have the platform instantly understand how to scale our API routes allowed us to focus entirely on squashing bugs and refining the business logic rather than configuring servers.

What's next for Quikka

Quikka is not just a hackathon project; it is the foundation for an unbanked financial ecosystem. Our immediate next step is to hit 200 active, daily power users across Kenyan university campuses by optimizing our viral loop. Technologically, while we currently use AI for product generation, we plan to hand over complete store management to AI. We will also leverage Vercel's backend AI infrastructure to build an automated arbitration engine that can instantly resolve buyer-seller disputes by analyzing WhatsApp chat logs and delivery photos.

Built With

• amazon-web-services
• aurora
• fastapi
• html
• javascript
• nextjs
• postgresql
• v0
• vercel