How Crazy Ant Labs sees and controls its team's AI

Crazy Ant Labs builds and invests in tools that make coding easier, for developers, by developers. Its products include SFTP To Go, managed SFTP and FTPS cloud storage, Cron To Go for scheduled tasks and Activity To Go for notifications when an app changes.

The company builds with a focus on scale, simplicity and security: other companies run their file transfers and jobs on its products. Its team is lean and fully remote, working from Israel, the United States and South Africa.

Its engineers use Cursor, Claude, Codex and Gemini, but the company didn't know exactly which AI tools were in use across the team.

The team was always concerned about staying compliant in the AI era, and about dev credentials and source code reaching an AI agent it couldn't control. It didn't want to control or monitor people's private accounts, but it wanted to make sure no company data leaked through them. With 100% of its development now done with AI, that risk is very high. What sent the team looking for a solution was its internal SOC 2 and HIPAA review, where AI security was the first requirement to answer.

Crazy Ant Labs rolled out Autonomous on more than a dozen endpoints over a few days. Within minutes of installing it, the team could see what people were using, including OpenClaw tests engineers were running on company machines that the team didn't know about.

The team spent a few days trying the product and getting to know it, running it in monitoring mode to see the violations and address them first, and then turned on enforcement. Preventing secrets from leaking and AI agents from reaching production access is one of the main policies it enforces.

The main value: the team ships faster and safer, and its customers get product updates and bug fixes faster than ever before. Everyone is happy.