Reinventing an MVP

Everyone has encountered someone passionate about creating an innovative product, whose vision transcends the notion of a Minimum Viable Product (MVP). In theory, an MVP involves building something as small as possible to solve a specific problem, addressing only essential needs. However, the reality is that creating something small may be easy to change, but scaling it can become a real challenge.

Generally, what people envision as a product is the evolution of an MVP after several iterations over time.

👶 My Start at UpdateAI

UpdateAI (UAI) is a tool designed to revolutionize the productivity of Customer Success Managers (CSMs) by incorporating AI to help users save time and enhance their results. Shortly after celebrating its first anniversary, UpdateAI had already surpassed multiple stages of experimentation. The lab had closed, and the MVP was in the final stages of development to go on sale. This is where my journey begins.

🤔 The MVP Dilemma

When UAI was finally ready to be introduced to the world, with Richard, the Engineering Manager, we faced a dilemma. While the UAI platform worked and could process orders, implementing changes and adding new features was a real challenge for us engineers. Each change took 2X to 3X the estimated time, and trusting the platform became difficult.

At that moment, we questioned whether the platform in operation could be considered a true MVP. We decided to start from scratch.

🏗️ UpdateAI 2.0

We zeroed in on what was essential for us to fully trust the platform’s capabilities and concluded that scalability was the key concept.

After several days of creating class diagrams, flowcharts, and establishing every possible use case, including the impossible ones (which were quickly discarded), we concluded that for the platform to shine, it must be scalable. We developed an execution plan, making scalability the core of the entire platform. This term applied not only to scaling the infrastructure on demand but also to ensuring that each new piece of code could be easily incorporated, and refactors should not be an impossible task.

After three challenging months of work — and several keyboard broken — we managed to replicate all existing functionality in a completely new platform, based on microservices, communicated through message queues and topics. Always adhering to best practices (clean code and use cases) and design patterns. We also developed a CI/CD pipeline for a smoother workflow.

Features were planned and executed within defined timelines, improvements were progressive, and it became much easier to identify business domains. Everything became a breeze, but we still didn’t know if it was the right path.

🧠 AIs

In May 2023, AIs took the stage of the entire tech industry, and there we were able to resolve our doubt: yes, we had taken the right path.

By envisioning the platform as a system of independent services with well-defined domains, when AIs entered the game, it became very easy for us to incorporate them to enhance our own AI developments. We could fine-tune iterations rapidly, allowing us to reach the product we are offering today.

🚀 Beyond the MVP

Today, it is challenging for me to consider UAI just an MVP. It has successfully processed over 3 million minutes of video and over 100 million webhooks from multiple platforms. Additionally, two central concepts were added to enhance the platform’s capabilities: security and traceability, enabling us to provide a better and secure service.

I am truly proud to be part of the founding fathers of this new platform. I have complete confidence that the foundations we laid are solid and robust, likely propelling the product to a new level. It is essential to highlight the human warmth and technical expertise of the entire team, without whom it would have been impossible to achieve.