Case Study: Scatter Lab achieves rapid proprietary LLM development with Databricks Mosaic AI

Delivering AI-powered entertainment that captivates users for hours

Scatter Lab, a conversational AI startup, wanted to build a proprietary LLM for its AI chatbots rather than rely on external APIs, but faced tight budget limits, limited GPU access, and the need for better data security, custom features, and reliable large-scale service. Its flagship products, including Iruda, Nutty, and Zeta, were growing quickly, making the challenge of scaling high-quality, human-like conversations even more urgent.

Using Databricks Mosaic AI, Scatter Lab built its own LLM in just three months. The result was rapid growth for Zeta, reaching 1.5 million users within nine months, with users spending more than 12 hours per week on the platform. Scatter Lab also reported 2 million downloads and 1.1 billion conversations across its chatbots, positioning the company for further expansion, including into Japan.

Scatter Lab

Jongyoun Kim

Chief Executive Officer