Cloud Reconciliation Agent — Eliminating Hidden Infrastructure Costs
Built an automated reconciliation system using Python and Temporal.io to align client sales references with the actual deployed cloud infrastructure. The agent detected unbilled or orphaned resources in real time, preventing multi-million-euro losses.

Snapshot outcomes
Revenue leakage prevented
€2M+ annually
Detection frequency
Daily sync cycles
Divergence resolution automation
100%
Problem
The reference data for client subscriptions often diverged from the actual state of the deployed cloud infrastructure. Some clients retained access to active compute and storage resources that were no longer listed in the sales system.
This mismatch led to substantial unbilled usage and inefficiencies in cost tracking, resulting in millions of euros in annual losses and heavy manual reconciliation work for the ops and billing teams.
Approach
Designed a Python-based reconciliation agent orchestrated with Temporal.io to automatically cross-check client references against live infrastructure states.
The system ran daily sync workflows that detected and flagged discrepancies, automatically classifying them as “orphaned,” “unbilled,” or “mismatched” resources.
Integrated alerting and reporting to internal dashboards so financial and ops teams could act immediately—or trigger automated cleanup and billing adjustments when safe to do so.
Impact
Eliminated recurring revenue leakage by automatically identifying and resolving infrastructure discrepancies before billing cycles closed.
Reduced manual auditing and data reconciliation to near-zero, freeing engineers to focus on proactive infrastructure optimization.
Recovered millions of euros in previously lost revenue and established a long-term safeguard against untracked cloud resource drift.
Want a similar sprint for your team?
Book a call and we’ll map out a 2-week plan tailored to your stack, your processes, and the outcomes you need right now.
Keep exploring
Case Study
Radar — Automated Worker Matching for On-Demand Staffing
Designed and shipped a feature that let the operations team automatically surface qualified worker profiles for last-minute missions in the food industry. The system replaced 50+ hours of manual search and outreach per week with an automated pipeline powered by Go, React, and Customer.io.
Read the storyCase Study
Automated Hours Validation System for Mission Tracking
Developed a smart import and validation system that automated the reconciliation of worker hours across client Excel files, database records, and worker confirmations. The AI-assisted process reduced weekly ops time from 50 hours to under 3 hours.
Read the story