Our team constantly seeks new ways to harness cutting-edge technology, pushing forward the decentralized and security ecosystems. That is why innovation and experimentation are at our core.
A recent creation that highlights this pursuit is Stuart AI. The project was developed over the past month by our security team and introduced to the rest of the company on our latest Office Days in Berlin.
Designed to improve code security, this security AI agent streamlines the detection of vulnerabilities in pull requests, ensuring safer code with minimal fuss. Stuart AI has been an invaluable exercise to help us build a product that enhances our internal processes and holds the potential to become a valuable tool for our clients.
The Power of AI Experimentation
AI is increasingly becoming essential in all areas of life and business, revolutionizing how we interact with technology and each other. Although AI, like all technologies, has its limitations, these challenges present valuable opportunities for growth and innovation. By pushing the boundaries of AI through experimentation, we contribute to its evolution, helping it mature and expand its capabilities.
Ivan Oudkerk, our Cyber Security Engineer, reflects on this process: “We are at a stage where AI, much like the early days of limited RAM or storage, faces its own set of constraints. However, I envision a future where AI can process vastly more information, making it increasingly accurate and effective at meeting our demands.”
We are committed to not only leveraging AI for its current benefits but also to exploring and expanding its potential. We want to ensure it continues to transform technology and business practices well into the future.
Origin of Stuart AI: Your AI Security Agent
Ivan, alongside Alejandro Acho, our Full Stack Developer, led this initiative. The project began from a simple yet critical need: to enhance the security of our code in an efficient and automated manner.
They recognized the potential of AI, especially LLMs, to revolutionize our code review processes. AI’s ability to understand context, identify patterns, and learn from interactions made it the perfect candidate for tackling this challenge.
Driven by the vision to integrate AI seamlessly into our development workflows, the project addressed current needs and set a foundation for future advancements.
The development process
Integrating sophisticated AI technology into our conventional workflows required a creative and methodical approach. The team faced the task of ensuring that the integration was seamless and could operate effectively within our existing infrastructure.
One of the most significant challenges was embedding LLMs into our development pipeline. This required them to continuously adapt and innovate, ensuring that Stuart AI could respond accurately to the needs of our codebase.
Ivan was pivotal in originating the idea for Stuart AI and setting up the project’s infrastructure. He also provided essential tools and prompts that proved invaluable in developing the MVP.
Meanwhile, Alejandro took the lead in designing and programming both the backend and frontend. He managed the integration of the application with LLMs and tailored the project’s requirements to the programming tasks. This ensured that every component worked seamlessly together.
Learnings and insights
The development of Stuart AI provided our team with profound insights and valuable lessons in the application of AI within our cybersecurity framework. This project advanced our technical capabilities and deepened our understanding of the practical impacts of AI on our workflows.
One of the key learnings was how to effectively integrate AI technology into existing products. It opened up new possibilities for using AI to empower rather than merely serve as a tool.
The team also learned to navigate the limitations of AI. Ivan’s insight into the need for precise prompt engineering to reduce hallucinations and improve the AI’s understanding of the code was crucial. This has led to a better grasp of how to fine-tune AI interactions to produce more accurate and reliable results.
Clovr Labs’ full commitment to innovation
The development of Stuart AI has profoundly influenced Clovr Labs’ approach to innovation. The project underscores the company’s commitment to innovation and cybersecurity, both crucial fields for technological advancement.
“Beyond just creating a product, using AI for this gave us a broader understanding of what we can do with LLMs to continue learning and innovating day by day,” highlights Ivan.
The next steps for Stuart AI involve refining the MVP for a public launch and potentialy open-sourcing it. We plan to collect user feedback to enhance vulnerability detection capabilities and fine-tune the model for even greater precision.
If you’re looking to drive your business forward innovatively and effectively, get in touch with us today. We want to help you transform your company and position you at the cutting edge of financial innovation.
Leave a Reply