Background
Reading code/discussions is a significant part of software development. As developers, we not only write code but also solve problems. To solve problems, we need to understand past solutions and the system we have to work with. Thus, reading documentation is required to understand previous discussions about similar problems, related systems, or even niche bugs.
Problem
Unfortunately, documentation in modern work environments is often scattered. Discussions often happen on multiple platforms like Gitlab issues, Slack channels, or private message rooms. In the worst of cases, documentation exists only inside the senior developer’s head. This problem gets worse as your codebase increases in complexity and more people get onboarded into the team. This practice is neither scalable nor sustainable, especially for companies whose product involves a complex domain where public discussions are rare.
How to solve them?
Developers need a way to find solutions to their development problems by digging through all the related platforms. This process can be time-consuming and frustrating. This problem is self-imposed, however, as finding relevant documentation will be far easier when the platform is centralized. For many companies, this is not a viable option, however, as discussions in a single platform might not be context-relevant (e.g., a discussion about a Gitlab repo might not be relevant to be put in a Slack channel). So, there must be a way to centralize all those scattered discussions into a single platform.
Our solution
Cephalink answers the problem statement above. Cephalink is a web application that integrates with your Gitlab and Slack accounts and helps you find the relevant documentation for your coding problems—as good as a Staff Software Engineer. Gone are the days where you had to spend hours poring through endless discussions and repositories just to find that bugfix you badly needed.
Cephalink provides a chatbot-like interface where you can ask any question about a specific problem in your code, and it will search through all Gitlab issues, merge requests, and commit messages and provide you with the recommended answer and several relevant links. It applies the retrieval process on a finite set of discussion embeddings. If you are not satisfied with the recommended answer, you can also conduct a deep dive yourself into the relevant links into the Issues/Messages provided.
Built With
- ai
- dx
- gitlab
- nextjs
- oauth2
- pgvector
- postgresql
- prisma
- tailwind
- trpc
- typescript

Log in or sign up for Devpost to join the conversation.