Rhythmic's AI Strategy Workshop Takes Blue360 From Early AI Experiments to a Shipped Platform in Six Weeks

Rhythmic’s AI Strategy Workshop Takes Blue360 From Early AI Experiments to a Shipped Platform in Six Weeks

Blue360’s developers were already experimenting with AI, and the company faced an ambitious platform rewrite on a tight deadline. It started as a workshop. Six weeks later they launched a working product and began selling it.

6 Weeks

From empty repository to live sales

20X Faster

Than the old manual editorial work

THE CUSTOMER

Blue360° Media is the leading U.S. provider of legal information exclusively for law enforcement. They offer a wide range of publications including manuals, code books, field guides, and professional development materials. Specializing in custom publications for specific agency needs, Blue360° Media’s mission is to inform and equip law enforcement professionals with the latest laws, rules, and regulations. Their focused approach ensures that those responsible for community safety have access to accurate and up-to-date legal information, supporting effective and compliant law enforcement practices.

THE CHALLENGE

Two things were happening at once. Like many engineering teams, Blue360’s developers had adopted AI faster than the organization had put standards around it. The work lived across sixteen separate tools and repositories, so no single place gave the AI, or the team, a complete view of the product. At the same time, the people who understood the product best, the product owners and the CEO, wanted a cleaner way to get their vision into developers’ hands without detail getting lost along the way. Sitting on top of both was an ambitious goal: a full platform rewrite, on a firm and fast-approaching deadline, that had to integrate with the existing e-commerce system the business already ran on. The team was moving fast against a hard date. What they needed was structure, shared standards, guardrails, and a reliable path from idea to shipped feature.

THE SOLUTION

It started as a workshop. Rhythmic came in to do two things: clean up the shadow AI usage and standardize the team on a governed pattern, and give product owners an AI-driven way to hand real requirements to developers instead of throwing work over the wall.

Because the team was small, three developers, two product owners, a project manager, and a few others, the workshop ran as individual sessions rather than one room. Rhythmic assessed where each person really was. How are you using AI today? What is plugged in, what is not? Then, one person at a time, the sessions showed the art of the possible on the team’s own code: live debugging, live fixes, prototypes built on the spot. Proving value in front of people, on their real work, is what earned executive buy-in. Once the C-suite saw it work, the conversation got easy, and the workshop turned into a build.

With value proven and leadership committed, Rhythmic put the guardrails in place. A CLAUDE.md and a set of developer standards defined one standard way to build and deploy. Rhythmic led a round of architectural consolidation, pulling work that had been spread across sixteen places into one standard so the entire product could be loaded into context every time the team worked with the model. Rhythmic’s Dobroslav Slavenskoj served as the team’s de facto architect for the duration.

Because Blue360 is an AWS shop, the team built an AI pull-request review system on Amazon Bedrock. Every merge request sends a condensed version of the change, plus the context it needs, to Claude via Bedrock for an adversarial review, and the feedback posts back into the request before anything merges. Re-commit to a work-in-progress request and another review fires. On a platform this size, spanning three databases and a multi-gigabyte dataset, the bug count stayed low and hotfixes were rare.

The most unusual move came from Blue360’s CEO. She had spent more than twenty years working with databases before becoming an executive, and the platform effectively lived in her head as a data model. Rather than describing features and waiting for someone to translate them, she used Claude to author the schema herself and hand finished schema files directly to the dev team. She became the project’s database designer. Once the design was finalized and live, everything after was additive. She now has her own pipeline into the platform for the editorial fixes only she catches.

The rewrite replaced Moodle, the team’s e-learning tool, with a training platform and content viewer Rhythmic built, and it integrated cleanly with the existing e-commerce system. After launch, Rhythmic helped Blue360 stand up single sign-on with a major metropolitan police department, an integration they had never been able to offer before.

THE RESULTS

Blue360 launched a working product on deadline and started selling it to existing and new customers almost immediately. From a bare repository to a tested, selling platform took about six weeks. A product owner told Rhythmic it was the fastest that team had ever shipped anything.

The change went past the launch. Manual editorial work that used to consume a month, like annotating six hundred pages of a book by hand, now moves roughly twenty times faster. Once executives saw the results, the same approach spread to functions that are not software development at all, from Salesforce administration to the editorial team, wherever a slow manual process was worth speeding up.

Product owners and developers are now pulling in one direction, with governed AI usage and a clear sense of where the tool helps and where it does not. People who had been stretched thin by the pace are getting time back, and stress on the team is coming down. Blue360 is out of the launch sprint and into steady, ongoing development.

TECHNOLOGY USED: