-
-
Linear + KodeLogs App Demo
-
repository list
-
config options and generate commits and translations
-
stakeholder, customer and developer tailored responses
-
stakeholder response example
Inspiration
Developers speak code; customers speak benefits, and shareholders speak profits. We saw a massive communication gap between technical teams and business stakeholders. KodeLogs was born to bridge that divide by turning complex code updates into clear, tailored changelogs for every audience.
What it does
Our AI translates technical code changes into understandable, audience-specific updates. For example, a terse developer note like "rAcE cOnDiTiOn ReSoLvEd" becomes a clear explanation such as "Fixed a timing issue that was causing users to get logged out unexpectedly." This helps developers, business teams, and customers stay aligned on what’s changing and why it matters.
How we built it We used Next.js and TypeScript with shadcn/ui for the frontend, Gemini AI for natural language translations, GitHub’s API to fetch code changes, MongoDB to store vector embeddings for context, and Linear Agents to maintain conversation flow. User authentication is handled securely via OAuth.
Challenges we ran into
One significant hurdle was securely accessing private repositories without directly handling users’ personal access tokens. Ensuring privacy and security while syncing GitHub changes required designing a robust authentication flow that could leverage OAuth securely and respect user permissions. Additionally, integrating real-time syncing of Linear issues with GitHub commits presented challenges in maintaining data consistency and handling rate limits from both APIs. Coordinating updates between the two platforms smoothly, without causing duplication or missed data, demanded careful design and thorough testing.
Accomplishments that we’re proud of
We delivered a working demo that takes a single code change and generates three perfect, audience-specific changelog explanations. Our exclusive Linear integration is fully operational, enhancing productivity.
What we learned
We realized that communication is never one-size-fits-all. We also discovered how critical prompt engineering and contextual vector embeddings are to making AI translations truly powerful.
What’s next for KodeLogs
We plan to expand integrations to Jira and other tools, offer customizable output formats, and eventually automate full CI/CD pipelines to generate changelogs seamlessly in real time.
Log in or sign up for Devpost to join the conversation.