Case Study: Dust boosts employee productivity and cuts vector costs by 2x with Qdrant

Using AI to Unlock Company Knowledge and Drive Employee Productivity

Dust, a company co-founded by a former OpenAI research engineer, faced the challenge of helping businesses leverage large language models (LLMs) to improve employee productivity. A major hurdle was tailoring these general models to a company's specific internal knowledge, which is difficult through fine-tuning. This led them to pursue retrieval augmented generation (RAG) to effectively utilize data from various SaaS applications, requiring a powerful and scalable vector database solution.

Dust implemented Qdrant as its vector database to build its platform of context-aware AI assistants. Using Qdrant Cloud on Google Cloud for a quick start, they later optimized performance and scaled by leveraging Qdrant's features like scalar quantization. This resulted in maintaining low latency and high accuracy across hundreds of thousands of collections while significantly reducing their infrastructure costs by 2x.

Dust

Stanislas Polu

Co-Founder