As developers, we all know the frustration of trying to find reliable answers to our coding questions. While tools like ChatGPT can be helpful, they often generate code with errors or hallucinations.
This is the problem I set out to solve with my startup Devv, an AI-powered search engine built specifically for developers.
Before writing a single line of code, I wanted to deeply understand the needs and pain points of our target users. I conducted 1:1 Zoom interviews with 50 developers from various backgrounds, digging into their current challenges and the solutions they were using.
A common theme emerged: developers were fed up with the inaccurate code snippets and explanations from generalist AI tools. They wanted a reliable, dev-focused alternative.
Armed with this key insight, we got to work building Devv. Our goal was to create an AI search engine that could provide trustworthy, error-free answers to coding questions and generate clean, functional code snippets.
Under the hood, Devv is powered by RAG (Retrieval-Augmented Generation) and large language models. To ensure the most relevant and accurate results, we built our own specialized search indexes spanning development-related websites, official documentation, and open-source code repositories.
To validate our idea quickly, we challenged ourselves to build and launch an MVP within just 1 week. The initial version was bare-bones but functional, focusing solely on solving the core problem of unreliable AI code generation and answers.
After shipping the MVP to a small group of friends and colleagues, the feedback was overwhelmingly positive. Developers loved having an AI tool they could actually trust for their day-to-day coding needs. This gave us the confidence to keep pushing forward.
We knew we weren't the only generative search engine out there. Competitors like Perplexity, You.com, and Phind had already established themselves in the market. While they also utilized Google or Bing search results for up-to-date information, we believed that to truly serve developers, we needed to go a step further.
That's why we made the decision to build our own custom search index specifically tailored to development-related scenarios. In addition to integrating Google search, we crawled and indexed a vast array of developer websites, official documentation, and open-source code repositories like GitHub.
This exclusive focus on developer resources allowed us to provide more relevant and accurate results for coding queries. Rather than relying solely on general web results, Devv could draw upon a rich knowledge base of programming-specific information to answer questions and generate code snippets.
To further differentiate Devv, we implemented two other key features:
Agent Mode: For complex queries, Devv infers the user's intent and selects from specialized agent types (code generation, explanation, debugging, optimization, web searches) to provide tailored solutions.
Clean, dev-focused UI: We designed Devv's interface to be clutter-free and intuitive, allowing developers to focus on their work without distractions from the tool itself.
By combining our custom developer search index with these unique capabilities, we were able to carve out a compelling value proposition for Devv as the go-to search engine for developers. Our specialized approach set us apart from the more general-purpose offerings of our competitors.
With an optimized MVP in hand, it was time to get Devv in front of a larger audience. We started by sharing it with my existing Twitter following, which consisted mainly of developers based in China.
The response exceeded our wildest expectations. Through word of mouth alone, Devv grew to over 500,000 users within just a few months. Developers across China were raving about the tool, and our user base was growing exponentially.
But just as we were hitting our stride, disaster struck. Devv was suddenly blocked by the Great Firewall, cutting off access for all of our users in mainland China. In an instant, our primary growth engine disappeared.
Losing our Chinese user base was a devastating blow, but we refused to give up. We knew we had built something of value, and we believed developers outside of China faced similar challenges.
So we went back to the drawing board, adapting Devv for a global audience. We started from scratch, rebuilding our user base one developer at a time. Slowly but surely, we began to gain traction in other markets.
As our global user base grew, we started thinking about monetization. We wanted to build a sustainable business without compromising the user experience.
In March, we launched a premium subscription plan for Devv. To our delight, it generated over $28,000 in revenue in the first month alone. And that number has only continued to grow in the months since.
Today, Devv is helping developers across the world write better code, faster. And we're just getting started.
Building and scaling Devv has been an incredible learning experience. Here are some of my key takeaways:
Solve a real, painful problem. The most successful startups are those that address a genuine need for their target users.
Validate quickly with an MVP. Don't get caught up in building the perfect product right away. Launch a bare-bones version to test your assumptions.
Differentiate your offering. In crowded markets, it's crucial to stand out through unique features and a laser focus on your target audience.
Word of mouth is powerful. If you build something people love, they'll tell their friends about it. This is especially true in tight-knit communities like developers.
Adversity is inevitable. No startup journey is without its challenges. The key is to stay adaptable and keep pushing forward.
At the end of the day, our success with Devv comes back to that initial insight: developers needed a better, more reliable AI search tool. By staying laser-focused on solving that core problem and continually improving our product, we've been able to build something that resonates deeply with our users.
If you're considering starting your own venture, I highly recommend checking out these valuable resources:
Y Combinator's Startup School: This free online program covers everything from ideation to fundraising, with video lectures from successful founders and investors. It's a crash course in entrepreneurship that I found incredibly helpful.
"The Lean Startup" by Eric Ries: This book is a must-read for anyone looking to build a startup. It emphasizes the importance of rapid experimentation, validated learning, and continuous improvement.
Lenny's Newsletter: Lenny Rachitsky, a former product leader at Airbnb, shares insightful essays and interviews on product management, growth, and scaling startups. His newsletter is a goldmine of practical advice and inspiration.
"The Hard Thing About Hard Things" by Ben Horowitz: Horowitz, a renowned venture capitalist, offers a raw and honest look at the challenges of running a startup. It's a sobering but essential read for any founder.
No amount of reading can fully prepare you for the rollercoaster ride of building a startup. The most important thing is to just get started and learn by doing. Start small, then iterate.
I hope my story inspires other founders to pursue their own startup dreams. If you're passionate about solving a real problem, don't be afraid to take the leap. The journey won't be easy, but the rewards are more than worth it.
Feel free to reach out if you have any other questions about my journey or need advice on getting started with your own venture. I'm always happy to chat with fellow entrepreneurs and share what I've learned along the way.
Very good share. How do you price your products?
ok great progress. one question? how was you able to scrape such a huge data of coding?
That is interesting to know
Why were you banned in china?
how do you plan to scale from now on ?
Hey Jiayuan, I saw your one time payment option for devv
does that convert better than the subscription you have in your pricing page?
Hi, nice journey, thanks for sharing.
Where did you find the initial 50 developers to talk to?
Hi, nice journey, thanks for sharing.
Where did you find the initial 50 developers to talk to?
Thanks for sharing such a highly insightful post! So many actionable tips and resources for beginners. Respect!
Love the detail in this post, will defo add to the next edition of smmry.tech.
hi, great story. And the tool works really nice, I've just tested it.
Could you give more details about the way you built your own index?
Did you use some SERP APIs? What was the cost in terms of compute?
What's on the roadmap in terms of development?
Do you plan to raise some money or will you keep it bootstrapped?