# How Zo Computer improved AI reliability 20x on Vercel

DevFeed: [How Zo Computer improved AI reliability 20x on Vercel](<https://devfeed.tech/articles/how-zo-computer-improved-ai-reliability-20x-on-vercel-757.md>)

Original publisher: [Read original article](<https://vercel.com/blog/how-zo-computer-improved-ai-reliability-20x-on-vercel>)

Author: Eric Dodds

Published: 2026-04-17T04:00:00Z

Content type: article

Language: en

Sources: [Vercel News](<https://devfeed.tech/sources/vercel-news.md>)

Topics: [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [backend](<https://devfeed.tech/tags/backend.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [latency](<https://devfeed.tech/tags/latency.md>), [models](<https://devfeed.tech/tags/models.md>), [openai](<https://devfeed.tech/tags/openai.md>), [routing](<https://devfeed.tech/tags/routing.md>), [vercel](<https://devfeed.tech/tags/vercel.md>)

## AI overview

Zo Computer describes moving its AI model-call infrastructure to Vercel, reporting lower retry rates, higher chat success, reduced P99 latency, and faster support for new models.

## Source excerpt

Zo Computer on Vercel 20x reduction in retry rate (7.5% -> 0.34%) 99.93% chat success rate (up from 98%) P99 latency cut 38% (131s -> 81s) New models added in less than 1 minute Every company has servers that store data, run services, and do work around the clock. Consumers just have apps. Rob Cheung, co-founder of Zo Computer, is closing that gap. Zo is a personal AI cloud: your own servers and data that power an always-on agent. "Cloud is one of the best computing models of all time, and consumers have zero direct access because it's so complicated," explained Rob Cheung, co-founder and CEO of Zo. "Now, with AI, it's finally possible for all of us to have cloud computers." Zo is a full computing environment, not just a chatbot. Rob laughs about his mom running servers and databases without knowing it. People use Zo to manage small businesses, do research, organize finances, and track health data. The 8-person company is two and a half years old and they have an ambitious goal: to onboard one million new users to personal cloud computing in 2026. That means millions of AI model calls every day, and when Zo users text their agent like a friend, they expect the same responsiveness. Death by a thousand adapters Zo gives users access to any model they want, and supports bring-your-own-key. That means their backend has to talk to every major provider: OpenAI, Anthropic, MiniMax, GLM, Fireworks, and more. Before they moved to Vercel, that meant custom adapter code for each model. Every provider required different handling for images, different key management, and different edge cases. On top of the code complexity, Zo's team was managing retries, provider routing, and fallback logic themselves. Every time a provider shipped a new model, an engineer had to write a new adapter, test the edge cases, and run the deployment pipeline. With new models released weekly, it was a constant drag on a small team building a consumer product, and their users felt it. Zo's baseline for AI