Gitconvo

Inspiration

We were building AI workflows and side projects but only when coding did we have access to Git version control. When either of us was designing prompt chains or conversations within AI chat windows like GPT or Claude we realized we had no versioning, no history, and no way to compare iterations. It was extremely annoying and cumbersome to not have any way to version control, save, and easily reference points within our chat and that’s when it hit us: prompt engineering and conversations need version control just like code. We wanted a tool to support AI workflows with real structure, collaboration, and memory.

What It Does

GitConvo lets users track AI conversations like software code. It introduces version control for LLM conversations, enabling teams to fork, compare, and build on AI-generated chats. Whether you’re working alone or collaborating with others, GitConvo creates a persistent history for your AI workflows, eliminating the chaos of screenshots, scattered chats, and lost context.

How We Built It

We built GitConvo using React for the frontend and Flask and Supabase for the backend, making it scalable and flexible. The tool integrates with OpenAI’s GPT-4 to begin with, and we’re planning to expand support to other LLMs like Claude and Gemini in the near future. We focused on creating a simple, intuitive UI for branching and collaboration, while also ensuring that users could store and revisit previous conversations. Version control was implemented to track changes across multiple threads of conversation, mimicking the way code changes are tracked.

Challenges We Ran Into

The biggest challenge was ensuring that the tool didn’t feel like a development environment. We had to strike a balance between advanced version control and making it accessible to non-developers. The second challenge was handling real-time collaboration without overwhelming the user interface with too many features.

Accomplishments That We're Proud Of

Building Git-style workflows that support branching, comparing, and remixing AI conversations.

Building wireframes for google-doc like real time collaboration and version history

Gathering early interest: users have already requested beta access based on the prototype demo.

What We Learned

Users who work with AI-generated content need structure, not just output. They want the ability to track changes, compare versions, and collaborate on their AI work just as they would on any other project.

Building for non-technical power users means providing advanced capabilities in a simple, intuitive interface.

What's Next for GitConvo

Onboarding more high-agency users who spend hours a day in ChatGPT, Claude, and Gemini.

Expanding our features to support real-time multiplayer workflows and more collaborative features.

Introducing integrations with platforms like Notion and Slack to bring GitConvo’s structured workflows into existing productivity ecosystems.

Scaling the platform for enterprise use, making it the go-to tool for AI-native teams.

Built With

• flask
• javascript
• openai
• python
• react
• supabase
• typescript