We Rewrote JSONata with AI in a Day, Saved $500K/Year | Reco

When Reco's engineering team faced mounting costs and performance issues with their JSONata implementation, they turned to an unconventional solution: rewriting the entire library using AI in a single day.

JSONata is a powerful query and transformation language for JSON data that's become essential for many SaaS applications. For Reco, it was critical infrastructure—but the existing implementation was becoming a liability. The team was dealing with high operational costs, slow performance, and maintenance headaches that were adding up to roughly $500,000 per year in total impact.

Rather than spending weeks or months on a traditional rewrite, Reco's engineers decided to leverage AI coding tools to tackle the problem. They fed the AI system detailed specifications of what JSONata needed to do, along with examples of the current implementation's behavior. The AI then generated a complete rewrite that maintained all existing functionality while addressing the core issues.

The results were immediate and dramatic. The new implementation ran significantly faster, reducing compute costs substantially. The cleaner codebase also meant fewer bugs and easier maintenance going forward. Most importantly, the entire rewrite process took just one day instead of the weeks or months a manual approach would have required.

This isn't just about saving money—though $500K annually is nothing to sneeze at. It's about how AI is changing the economics of software development. Tasks that once required large teams and extensive timelines can now be accomplished by small groups in a fraction of the time. For startups and growing companies, this kind of efficiency can be the difference between success and failure.

The JSONata rewrite is part of a broader trend Reco is seeing across their operations. As they've scaled their SaaS security platform, AI has become an increasingly important tool for handling everything from code generation to security analysis. The company recently announced AI agent visibility features that help organizations discover and manage the thousands of AI agents operating across their SaaS environments—agents that create toxic permission combinations traditional tools can't detect.

For Reco's customers, these efficiency gains translate directly into better security outcomes. When the company's own engineering team can move faster and more cost-effectively, they can invest more resources in building features that help organizations close the context gap in security investigations. Their recent Agent-to-Agent workflow with Torq, for instance, can autonomously investigate potential insider threats in seconds rather than hours.

The JSONata rewrite demonstrates something important about the current state of AI in software development: we're moving beyond experimentation into practical, high-impact applications. This wasn't a toy project or a proof of concept—it was production code that immediately started saving the company significant money while improving their product.

As AI coding tools continue to mature, we can expect to see more stories like Reco's. Companies that learn to effectively leverage these tools will have a significant advantage in terms of development speed, cost efficiency, and ultimately the quality of their products. The $500K annual savings is just the beginning—the real value is in what that freed-up capital and engineering time can now be used for.

For other engineering teams considering similar approaches, Reco's experience suggests that the key is having clear specifications and well-defined requirements. The AI tools worked because the team knew exactly what they needed the rewritten code to accomplish. This isn't about replacing engineers with AI—it's about giving skilled developers tools that let them work at a speed and scale that wasn't possible before.

The JSONata rewrite is a concrete example of how AI is reshaping software development economics. When a one-day project can save $500K per year, it changes how companies think about technical debt, performance optimization, and the allocation of engineering resources. It's not just faster development—it's fundamentally different economics.