About

Unfetch.com is an online platform designed to generate, deploy, and run AI agents that automate tasks. Acting as an Integrated Development Environment (IDE), it combines coding with an online deployment platform specifically tailored for AI applications. Unfetch agents are compatible with tools like the OpenAI GPT Store and LangChain, making code generation and automation seamless. The platform enables users to create AI workflows and interact with any existing API using large language models, simplifying the development of agents and API connectors.

How they use Cloudflare

For fast performance, Unfetch.com deploys Nuxt application through Cloudflare Pages, which has by and large been the team’s best deployment experience. Unfetch uses Pages Functions to handle essential tasks, such as processing logins, payments, and billings, ensuring smooth operations.

D1 is key to their data management. Users can define schemas for accessible, filterable, and searchable information. Whenever a file is uploaded, it's processed by large language models (LLMs) to extract data, which is then saved. When a customer requests data, it's quickly fetched via D1. Unfetch.com plans to implement a separate database for each customer once that feature goes live.

For larger files, they rely on Cloudflare R2 for storage. All types of documents are stored and accessed via the platform through R2, ensuring scalability and reliability.

Finally, AI Gateway serves as their safety net, flagging inference request errors while logging all AI requests across inference providers. It helps with caching, making their AI interactions smoother. Looking ahead, Unfetch.com plans to use Cloudflare Queues to process documents asynchronously on the client side, further enhancing their platform’s efficiency.

Why Cloudflare?

"Cloudflare Pages is incredibly fast—you deploy it, and it just works. The site loads instantly, and caching is seamless. I also appreciate the ability to rate-limit requests to specific endpoints, which is especially important when working with large language models (LLMs). It helps avoid potential massive bills if something goes wrong.

D1, when paired with Drizzle, performs beautifully. You can connect locally via API, and in production, the connections through bindings are near-instant. With replication on the way, it’s only going to get faster. Plus, Cloudflare’s pricing is unbeatable—often 10 times more affordable than the competition. As part of the Workers Launchpad Cohort #3 and a long-time user, I find Cloudflare delivers exceptional value without compromise." Constantin Bignoli, CEO