The Protocol & Technology

At ParlayTix, we've built something different. While most ticketing platforms rely on monolithic architectures that become bottlenecks under load, we've embraced a distributed microservices approach inspired by modern cloud-native principles.

Our system isn't just about finding tickets—it's about orchestrating intelligent agents that work together like a well-coordinated team. Each service has a specific role, a clear boundary, and communicates through well-defined contracts. This separation of concerns allows us to:

• Scale individual components based on real-time demand
• Deploy updates without system-wide downtime
• Maintain 99.9% uptime even when external ticket sites fail
• Process 10,000+ concurrent requests without degradation

Traditional ticket platforms try to do everything in one place—chat, search, automation, and data processing all tangled together. This creates fragility. When one feature fails, everything fails.

We took a different approach. By separating our system into five specialized services, each optimized for its specific task, we achieve remarkable efficiency. The Frontend Agent excels at understanding human conversation. The Agent Swarm orchestrates complex multi-step operations. The Crawler Core navigates hostile web environments with sophisticated anti-detection. Each service is a master of its domain.

The magic happens in the communication layer. Our services don't just pass messages—they engage in intelligent conversations. When you ask Pepper about Lakers tickets, your request triggers a carefully choreographed sequence that feels instantaneous to you but involves sophisticated orchestration behind the scenes.

The Frontend Agent immediately acknowledges your request, maintaining the conversation flow while the Agent Swarm begins its work behind the scenes. If fresh data is needed, a crawl request is published to our RabbitMQ message bus—not as a simple command, but as a rich context object containing your preferences, budget constraints, and timing requirements.

The Crawler Core, constantly monitoring the message bus, picks up the request and spins up a browser session through BrowserBase. Using Playwright with custom anti-detection measures, it navigates ticket sites as a human would—clicking buttons, waiting for content to load, even solving simple challenges. The scraped data flows back through the message bus, gets processed and stored by the Data API, and ultimately returns to you as a conversational response from Pepper.

This asynchronous architecture means you're never waiting for slow external sites. You get immediate feedback, and detailed results arrive as they become available. It's the difference between a frustrating wait and a delightful conversation.