Dormant cloud resources waste 35% of cloud spend — idle instances, orphaned volumes, forgotten dev environments left running. AI optimization recommendations catch this waste continuously, not in a one-off spreadsheet exercise nobody acts on.

Without a structured workflow, optimization recommendations die in inboxes: nobody owns them, nobody tracks the savings, and costs creep back within a quarter.

Utilyx models IT cost optimization as an executable decision tree — data collection, AI analysis, prioritized recommendations, action tracking and verified savings reporting.

The workflow is structured as a decision tree with 4 main branches (Usage Analysis, AI Recommendations, Action Tracking, Reporting) running on a scheduled cadence (weekly or monthly), with parallel action sub-branches and 8 steps end-to-end. Every step is executed by a named actor (AI Copilot, ai-agent, IT Manager, Finance) and every action is appended to the timeline_optimization audit subtable.

• Step 1 (Start — sql-query usage analysis): the workflow pulls cloud billing and usage data via sql-query into the usage_analysis table with fields resource_id, resource_type (instance / volume / s3 / lambda), monthly_cost, cpu_utilization, memory_utilization, last_activity_date, provider_account.

• Step 2 (Dormant license detection — sql-query): the workflow cross-references installed software against usage signals (login, launch, network activity) via sql-query to flag licenses paid but unused — writing to the dormant_license table with fields title, edition, install_count, last_used_date, wasted_cost.

• Step 3 (Hardware utilization review — sql-query): CPU, memory and storage utilization from discovery telemetry are analyzed via sql-query to flag over-provisioned servers and under-used hardware in the hardware_utilization table.

• Step 4 (AI cost-reduction recommendations — ai-agent): the ai-agent synthesizes all three analyses and prioritizes recommendations by estimated_savings and effort (rightsizing / termination / reallocation / consolidation) — recorded in the cost_recommendation table. Sub-branch: the tree splits on estimated_savings — Branch A (estimated_savings ≥ €1K/month): high priority, route to IT manager for immediate action; Branch B (estimated_savings < €1K/month): log for next quarterly review.

• Step 5 (Action tracking): each recommendation becomes a tracked task via sql-query on the optimization_action table with fields action_id, recommendation_id, owner, deadline, realized_savings, status — visible on the dashboard.

• Step 6 (Savings verification — ai-agent): the ai-agent compares projected_savings vs realized_savings on closed actions and flags variances > 20% for review.

• Step 7 (usePDF optimization report): usePDF generates an optimization report with findings, recommendations, realized and projected savings — for finance and leadership, in 7 languages.

• Step 8 (useArchive): useArchive retains every analysis and savings record for the financial audit trail with index fields analysis_date, recommendation_id, realized_savings, resource_type.

Organizations using Utilyx cut the 35% dormant cloud waste to under 5% (ai-agent flags idle resources on every cycle), achieve 15-25% total IT spend reduction in the first year (cloud rightsizing, license termination, hardware consolidation), with tracked and verified savings — not one-off estimates.

The no-code designer and AI copilot let an IT manager build the cost-optimization workflow in one session — no consultants, at 70% lower TCO than legacy FinOps and ITAM suites. Savings are verified monthly, so costs do not creep back.