Case Study: Notion cuts AI latency 4x and scales enterprise workflows with Fireworks AI

How Notion Cuts Latency 4x and Scales Enterprise AI Workflows with Fireworks AI

Notion, a productivity software company with over 100 million users, faced the challenge of scaling its AI capabilities beyond simple chat to power sophisticated, enterprise-grade agentic workflows. The company needed its AI to integrate reliably with tools like Slack and Jira to create a seamless user experience, all while overcoming significant latency, cost, and reliability hurdles. Partnering with Fireworks AI, Notion sought a solution that would allow its engineers to build and deploy efficient AI at scale.

By utilizing Fireworks AI to fine-tune smaller, more efficient models, Notion achieved a dramatic reduction in latency from two seconds to just 350 milliseconds. This 4x improvement in speed enabled the company to reliably serve its massive user base with low-latency AI workflows. The solution provided by Fireworks AI empowered Notion to rapidly iterate on features and successfully launch its vision of intuitive "vibe working" at an enterprise scale.

Notion

Sarah Sachs

Head of AI Engineering