Quinn Slack co-founded Sourcegraph to address the challenges his team faced in their own projects. After simplifying their internal process for navigating complex codebases, they realized that while smaller companies might not see this as a pressing issue, many enterprises did.
Sourcegraph is the leading code intelligence platform that revolutionizes how developers understand, fix, and automate their code. By leveraging AI through Cody, their AI coding assistant, Sourcegraph enables developers to easily navigate large codebases, locate relevant code snippets, and gain historical context—all while streamlining bug fixes, code refactoring, and performance improvements in a single interface.
Don’t miss this episode of Barrchives where Quinn discusses the evolution of Sourcegraph, the critical role of context in coding, and how AI is set to transform the role of the developer.
Watch it right here:
Below are 5 insightful takeaways from my conversation with Sourcregraph’s CEO Quinn Slack:
In Quinn's Words:
“That three-person company that’s been around for a few months. Yeah, they don’t have a lot of code. Maybe they don’t have as big of a problem as Bank of America with a big code base. But if they have 5% of the problem, then 5% of the time we’re going to win them as a customer, or we can charge 5% as much money. And it turns out that is not true whatsoever. You you can only go and convince companies that have a certain size and a certain intensity of that problem.”
In the early days, Sourcegraph targeted small companies like themselves that seemed like a perfect fit for their code search solution. But Quinn realized many of those small startups didn’t have the complexity in their codebases to feel the urgent need for robust code search tools. This miscalculation forced them to pivot toward larger enterprises, where the demand for effective solutions was much stronger. This shift not only allowed them to find their PMF but also became their edge and engine for success today.
In Quinn's Words:
“If you believe something, you got to go with it because you're not going to do a good job of executing on something that's antithetical to what you believe.”
Quinn encountered external pressures and conflicting advice that didn't align with the insights he gathered from customers, which created some tension but also underscored the need to trust his instincts. Recognizing the demand for a self-hosted solution based on customer feedback about security marked a turning point for Sourcegraph's growth. By prioritizing customer insights over outside opinions, Quinn solidified their strategy and forged partnerships with major players like Uber. This experience highlights the importance of staying true to one’s vision while being responsive to user needs.
In Quinn's Words:
“What AI did is it let us hold their hand a few more steps along the way so that we're not just showing them where in the code it's used, but actually in natural language, what might break.”
While AI can generate simple applications, its real value lies in understanding the intricacies of existing code. Sourcegraph leverages its experience to provide insights that enhance collaboration between human intuition and AI capabilities. By helping users identify not just where code is used, but also potential pitfalls, Sourcegraph empowers developers to make informed decisions. This highlights the need for tools that offer comprehensive context, reducing errors and improving productivity. As coding becomes increasingly complex, the tools we use must enhance collaboration, bridging the gap between human developers and AI.
In Quinn's Words:
“The hardest thing is figuring out when we can use ‘magic’ ... We set the expectation that you could ask anything and we would just magically find the right context. And look, we are better at context than any other code AI tool. But we're not magic.”
Quinn explores the tricky balance between delivering a "magical" user experience and ensuring accuracy. Initially, Cody Chat allowed users to ask open-ended questions, but this flexibility led to inconsistencies in the AI’s responses. To address this challenge, Sourcegraph implemented a more structured interface that used explicit at-mention chips, guiding users in their input. While reducing the "magic" of open-ended interactions may seem like a drawback, it builds trust among users by prioritizing accurate results. Quinn notes that taking an extra 30 seconds to specify input can save users 30 minutes in the long run.
In Quinn's Words:
“I think they [developer tools] become headless and they lose a lot of their pricing power and prominence in the developer toolkit because no human is going to actually know what tool they use under the hood.”
As AI becomes more integrated into the development process, Quinn prompts us to consider how these tools will evolve to meet the needs of an AI-centric environment.
Developer tools will need to streamline their functions and operate more seamlessly behind the scenes. This means that developers may find themselves using sophisticated tools without really understanding their inner workings. While this can pose challenges—such as a lack of control or insight—it also opens up exciting opportunities for innovation and efficiency. This emphasizes the need for developers to rethink how they approach tool design. The focus will shift towards creating tools that prioritize integration and efficiency, enabling developers to leverage AI capabilities without getting bogged down by complexity.
.svg)