At Jest, we are a distribution company at our core. We don’t “do marketing” - we solve for growth by merging engineering, product, and marketing into a single functional unit. We treat the entire user journey as a technical stack rather than a series of disconnected ads.
In this model, testing drives performance. Finding winning creatives and lowering acquisition costs requires running hundreds of tests simultaneously. However, manual campaign management is a bottleneck: standard UIs are sluggish, human error is inevitable, and talent is wasted on repetitive setup.
Previously, this friction was an unavoidable cost of doing business. Today, AI has redefined what is possible, allowing us to build a truly autonomous solution: a distribution agent we call HypeCommand.
Meet HypeCommand: Our Growth Agent
Instead of relying on traditional manual oversight to manage the chaos of Meta Ads Manager, we built a machine to do it for us.
HypeCommand is a custom agent designed to handle the heavy lifting of our distribution logistics. We automated our entire UA testing lifecycle by turning it into a programmatic workflow.
To run a test for a new app, our team simply drops assets into a local project folder and tweaks a configuration file. From there, HypeCommand takes over - programmatically validating the campaign, verifying assets against Meta’s servers to prevent duplicates, structuring ad sets, routing custom links, and deploying the ads nearly instantly.
The AI Accelerant
Building custom marketing infrastructure from scratch used to be a significant undertaking. The Meta Marketing API is dense and documentation is often outdated or missing. Building a robust integration can easily sideline an engineer for weeks.
This is where AI became our ultimate accelerant. By leaning heavily on AI for code generation and API navigation, our team went from an empty codebase to a fully functional Meta integration - and a working version of HypeCommand - in 24 hours.
Under the Hood: The Infrastructure
HypeCommand is a custom Node.js agent that interacts directly with the Meta Marketing API. Instead of fighting with a sluggish web interface or brittle browser macros, we turned our entire User Acquisition testing workflow into executable code.
Here is how HypeCommand operates in practice:
- Folder-Driven Deployment: Our growth team organizes all the new image and video assets into structured folders. The hierarchy of those folders mirrors the exact Campaign and Ad Set structure we want to create in Meta.
- Simple Configuration: A basic
config.yamlfile in each folder tells HypeCommand exactly where to route the new destination and tracking links. - Smart Reconnaissance: When a team member runs a command, the agent does not blindly upload files. It first audits the referenced folders and checks the media against Meta’s servers. If an asset was used in a previous test, it intelligently reuses the existing media. This drastically reduces upload times and avoids API rate limits.
- Programmatic Execution: With the assets verified, HypeCommand takes over completely. It duplicates the necessary ad sets, updates all the URLs, and publishes everything directly into Meta Ads Manager.
Scale Without Limits
What used to require up to several hours of clicking, waiting on loading screens, dealing with ad network UI bugs, and checking for human errors is now executed with a single terminal command.
That is how we were able to reach our current scale. Over the last few months on Meta alone, HypeCommand has ran over 5,000 growth tests.
It simply would not have been possible without a programmatic solution.
Growth today is not just about having the best creative ideas; the real edge is technical infrastructure. You need systems that let you run more tests faster with less manual effort and far fewer mistakes. By viewing growth through an engineering lens and using AI to build the tooling we need, we are building a distribution engine that can scale without limits.