Introduction

Perplexity AI is a startup that serves a best-in-class AI-powered answer engine. At AWS re:Invent 2023, Perplexity announced the availability of their online AI models, which are connected to their search index for factual grounding. In this case study, Dmitry Pervukhin and Aarash Heydari, members of technical staff specializing in infrastructure at Perplexity, share insights into why they chose Octane to power the usage-based pricing subsystem of their public self-service API.

‍

Challenge

Perplexity released the beta version of their API in early October. Their API was freely available to pro subscribers under restricted rate limits. To relax these limits and make their API generally available, Perplexity needed a reliable system for charging users based on usage. Pervukhin shares his thought process on the challenges of building a billing system.

“As a former research and development lead at an internet service provider and system developer for multiple credit card processing systems, I understand the place of a billing and payment system in a company's architecture. Billing systems are incredibly complex and have a huge number of corner cases. After proposing the simplest possible design, it became clear that it would take too long to build, be tedious to maintain, and probably would not deliver the feature set required for the operators. As a focused and fast-moving startup with limited resources, this was not acceptable.” 

Pervukhin continues, “From their first demo, it was clear that Octane had solved these challenges in a scalable and flexible way, offering simple and flexible abstractions for usage-based pricing. By following their documentation, it was easy to implement the majority of the integration without much back-and-forth with the Octane team. With their help, we were able to roll out usage-based billing within one month of our first conversation with them, and are now happily focused on the problems of our business instead of this supplemental system.”

Heydari reflects, “Octane’s team was nimble and helpful on a short schedule to help ensure that all the stakeholders at Perplexity got what they needed out of our billing solution. The integration with our existing system was rather seamless, and we have peace of mind that it can handle orders of magnitude in growth of our scale. I am also confident that the team at Octane will be there to help us resolve any issues we face along the way.”

‍

Solution

Octane swiftly addressed Perplexity AI's challenges, delivering a comprehensive billing solution. The key interventions provided by Octane included:

1. Rapid Delivery of Billing System: Octane delivered an end-to-end billing system within a remarkable 4 weeks, ensuring Perplexity AI could meet their API launch deadline.
2. Real-time Credit Transactions and Automated Top-offs: Octane's solution supported real-time credit transactions. Pervukhin emphasizes, "Automated top-offs and pre-purchased credits are important features for fraud detection and user experience which would have been painful to implement ourselves.”
3. Deep Integration with Stripe: Octane seamlessly integrates with Stripe. This integration allowed Perplexity AI to continue using their existing payment system while gaining the benefits of Octane for usage-based API billing.
4. User Visibility and Dashboards: Perplexity AI was able to build an in-app UI portal for purchasing credits while leveraging Octane’s end-customer portal to give users rich insights into their own usage. Heydari states, “Users were asking us for CSV reports of their usage. Octane was able to deliver this and much more. The best part is that our product team had an easy time integrating Octane’s portal into our frontend."

‍

‍

Use Cases

The implementation of Octane at Perplexity AI yielded significant benefits:

• Time and Resource Savings: Perplexity AI saved significant engineering time and resources by using Octane instead of building a billing system in-house.
• Enhanced Security and User Experience: Real-time credit transactions and automated top-offs added security layers and improved the overall user experience.
• Seamless Integration: Octane's integration with Stripe seamlessly extended Perplexity AI's billing capabilities without requiring interfering with the existing payment layer.
• End user experience: Octane’s embedded dashboards empowered Perplexity AI with an out-of-the-box solution allowing customers to visualize their usage with minimal work required from Perplexity AI’s engineering team.

In conclusion, Perplexity AI's collaboration with Octane showcases the pricing and billing capabilities that were unlocked for Perplexity AI through Octane’s out of the box solution. By using Octane, Perplexity AI was able to meet tight deadlines, enhance the overall user billing experiences, and seamlessly integrate with existing systems. This case study serves as an example of innovation driving efficiency in the dynamic realm of AI-powered services.