Faire's AI Agents Double PR Output

E-commerce company Faire has significantly boosted its engineering velocity, doubling weekly pull request throughput and drastically shrinking a major migration project. The company achieved this by replacing its in-house background agent system with Cursor Cloud Agents, a move that consolidates its approach to agentic development.

This shift allows Faire to leverage scaled parallelism and agent autonomy, overcoming the resource constraints of local development environments. Each cloud agent operates with its own dedicated environment, akin to a human engineer, enabling it to write, test, and verify code independently.

Scaling Beyond Local Constraints

Running multiple agents on a single machine quickly depletes local resources and becomes difficult to manage. Faire previously attempted to build its own solution, dubbed 'Samurai,' on self-hosted infrastructure, but found the investment in talent and maintenance prohibitive.

"Standing up our own servers is a significant investment. We’d rather have engineers focused on adding value to our end users," said Luke Bjerring, principal engineer at Faire.

The company selected Cursor for its deployment flexibility, GitHub integration, reliability, and streamlined user experience for managing concurrent agents. This makes Cursor the company's recommended agentic development platform.

Autonomous Agents with Dedicated Environments

The effectiveness of parallelized agents hinges on their ability to function like engineers: managing dependencies, interacting with internal services, and validating changes. Without a properly configured development environment, agents can write code but cannot complete tasks end-to-end.

Faire utilizes Cursor’s agent-led onboarding to navigate its complex development setup, which involves separate repositories for backend and frontend with distinct build tools like Gradle and Bazel, and separate AWS credentials. Cursor inspects each repository to generate environment configurations that can be versioned and edited.

"We let Cursor onboard itself on every repo in our codebase. That takes a lot of the overhead out of new session starts and lets agents tackle tasks just like an engineer would," noted Blair McAlpine, senior engineer at Faire.

This setup enables agents to power features like generating React components from Figma designs and producing demo videos, or directly addressing bug reports and code review requests originating from Slack channels.

"A lot of our work comes from ideas and discussions in Slack. You can see the message, kick off @cursor in the same context, and you get a PR a few minutes later," Bjerring added.

Automating Repetitive Tasks with Cursor Automations

Beyond parallel execution, Faire employs Cursor Automations to handle over 2,000 autonomous agent runs weekly. These automations streamline tasks such as triaging bug reports, automatically fixing CI failures, and routing code reviews.

"The concept of automations has been long-lived at Faire, but setting them up was painful and complicated. Cursor Automations makes spinning up always-on agents accessible to every user," said McAlpine.

Accelerating Legacy Migrations and Build Previews

Cursor's capabilities were instrumental in migrating a large application from MobX to native React state management. Using an agent coordination system called 'Swarm' built on Cursor, Faire delegated migration tasks to individual cloud agents. What was estimated to take 18 months of manual work for an entire team was condensed to a single engineer overseeing a fleet of agents.

The platform also enabled the rapid creation of a build preview tool. Previously a weeks-long endeavor, an engineer used Cursor to plan and execute the build, with a cloud agent delivering working preview builds in less than a day.

"The cloud agent ran in the background while I worked on other things. It took the preview builds from scratch to a working internal tool in less than a day," McAlpine stated.

Faire is now exploring how to extend this leverage to adjacent teams, aiming to unlock further momentum across the broader product development process and reallocate engineering resources to more ambitious projects.