The founders argue that while GitHub and OpenAI currently have an advantage due to their ability to deploy large AI models via the cloud, this edge will diminish as other AI models improve and computing costs decrease. TabbyML aims to lower the deployment barrier by recommending models trained on fewer parameters, which may result in lower quality in the short term but is expected to improve over time.
Key takeaways:
- TabbyML, an open source code generator built by two ex-Googlers, has secured $3.2 million in seed funding. It is seen as a customizable alternative to GitHub’s Copilot.
- Co-founder Meng Zhang believes that open source software meets the needs of larger enterprises better than proprietary software, and that all companies will eventually need some sort of customization in software development.
- Despite the potential for bugs, code generators are seen as a tool to assist programmers rather than replace them. For example, GitHub's Copilot users accepted 30% of the suggestions generated by the coding assistant.
- Zhang argues that the advantage of GitHub and OpenAI will diminish over time as other AI models become more powerful and the costs of computing power decrease.