RESULTS
10X
cost savings with AI-native approach
35
risks identified and quantified before a single line of code changed
10
guardrails established to protect the most sensitive code from accidental changes
Executive Summary
With its HTML5 Photo Editor enabling significant revenue, Storio Group saw an opportunity to make the product faster and safer to evolve. Storio partnered with EPAM to build a structured agentic AI workflow, using Claude. In weeks, the team made the codebase legible, safe and predictable to maintain. The result: a clear path to innovation at a fraction of the cost of a conventional rebuild.
CLIENT
Storio Group
SERVICES
- Artificial Intelligence
- Generative AI
STRATEGIC PARTNER
- Anthropic
INDUSTRY
- Consumer
Human-AI Partnership is Key to Success
As a European leader in personalized photo products, Storio Group’s HTML5 Photo Editor is a marquee product and earner. The company wanted to unlock revenue growth by rethinking how the product gets built and maintained. So Storio chose the AI-native path. Instead of tying up engineers in a slow, high-cost overhaul, the company wanted to apply agentic AI to accelerate the work, with governance framed around a simple, honest premise: AI will sometimes be wrong, so we gate for it. That discipline turned speed into an advantage rather than a liability, giving Storio the confidence to move quickly on a product it can’t afford to let stand still.
Storio partnered with EPAM to produce measurable value quickly. EPAM built a structured agentic workflow, using multiple Claude models powered by Anthropic.
As a result, a 13-year-old complex legacy codebase with minimal internal knowledge was transformed to enable safe, efficient change in a core company category.
First, we made the code legible by mapping and documenting all modules, dependencies and risks. Then, we created a cross-referenced knowledge base with evidence-tagged documents and mapped off-limits zones to prevent dangerous edits.
Next, we deployed three specialized agents, each with a single job that checks the others:
The Analyst
plans the change and writes the tests it needs
The Implementer
builds it test-first, re-checking every safety gate
The Validator
verifies the full change and signs off or sends it back
A human reviews and approves throughout, and every change must pass automatic checks for style, types and tests before it can be committed. As a result, the inconsistent test suite was stabilized, real regressions were caught and traced to their source, a third-party dependency was cleanly removed and the missing test-coverage baseline was added. The knowledge base self-heals with each task and regenerates from the codebase itself, so any engineer can now contribute with confidence.
The payoff is both financial and strategic: lower cost to modernize, faster time to maintainable product and a team that spends its energy on innovation instead of maintenance.
LEARN MORE
EPAM's AI-native engineering expertise helps modernize systems for better security, predictability and scalability.