Rabata.io decreased software costs by 15% using Subscope by detecting unused software and optimising seats
Challenge
Rabata.io — S3-compatible secure cloud storage and HLS video encoding platform.
Rabata is cloud infrastructure for product teams: S3-compatible storage and video in one stack, at a fraction of hyperscaler costs. Rabata controls the full infrastructure layer, storage, transcoding, and CDN delivery, so costs stay predictable at any scale.
A team of over 30 engineers and support specialists powers these operations.
AI and software are employed to handle a wide range of day-to-day tasks, helping to optimize every aspect of the workflow.
However, as the company grew, issues arose regarding the uncontrolled proliferation of software tools: for instance, subscriptions used by former employees remained active, and it became difficult to track monthly expenses or determine which subscriptions were actually in use. There were numerous instances where we paid for annual subscriptions and only cancelled them after the next payment for an unused service had already gone through.
Approach
We started to use Subscope because it has two game-changer features: Zero installation time and free to start. For us was crucial to see results before we pay anything upfront.
The UI was clear but what was even better was Subscope messaging system. We added the whole team in a few minutes and started to get notifications on upcoming payments, so we could react before payment happens. Then we used AI-tool to manage software actions such as quick employee addition or removal.
Results we achieved instantly:
- 47% of async handlers returned before their internal operations finished
- Several “harmless” utility functions introduced hidden parallel execution
- Error propagation was inconsistent across 6 core services
- One legacy module triggered three parallel writes without guaranteeing order
Using these insights, we built a prioritized repair roadmap:
- Resolve non-awaited critical paths
- Normalize error propagation
- Untangle chained conditionals into clear async sequences
- Stabilize flows powering the redesigned components
Solution
Within 5 weeks, Evermind shipped:
- A fully mapped async architecture across their 9 services
- Refactored logic in the top 14 error-prone modules
- Rewritten utility functions to enforce consistent awaiting
- Prediction of high-risk redesign components before they were built
- A unified code reasoning layer that prevented regressions during rollout
After stabilizing the async foundation, the platform redesign moved forward three times faster.
Results
What’s next
With the redesign shipped successfully and async issues under control, Evermind is now expanding Nerdstack to:
- monitor new modules proactively,
- onboard new engineers with automatic code explanations,
- and maintain long-term architectural clarity through continuous Review.
Their next milestone:
full system-wide observability of async flows by Q4 2026, ensuring the redesign becomes the foundation—not another layer of technical debt.
