AgentHack submission type
Agentic Testing Solutions
Name
Marco Zappa
Team name
Agents00Emissions
Team members
marco.zappa@avvale.com,simone.coslovich@avvale.com,davide.magnaghi@avvale.com,davide.giudetti@avvale.com
How many agents do you use
Multiple agents
Industry category in which use case would best fit in (Select up to 2 industries)
Compliance
Complexity level
Advanced
Summary (abstract)
An AI-powered testing pipeline that slashes CO₂ emissions, run-time and audit effort—by executing only what matters, when it matters. Built in UiPath, it dynamically adapts to change and delivers carbon-smart, audit-ready test reports across multiple departments
Detailed problem statement
Traditional test pipelines trigger full test suites on every code commit, regardless of the grid conditions—leading to long runtimes, high cloud costs, and significant CO₂ emissions. At the same time, compliance reporting is slow and manual, requiring teams to spend hours compiling logs, screenshots, and traceability artifacts. These inefficiencies scale poorly across departments, limiting agility and sustainability in regulated environments.
Detailed solution
We developed a low-code, agentic test pipeline entirely within UiPath, orchestrated using Agent Builder, Test Suite, and Maestro. The architecture includes two autonomous agents and supporting components that coordinate test execution and audit reporting based on carbon efficiency.
-
Carbon-Aware Scheduling Agent
An unattended bot runs hourly and retrieves up-to-date CO₂ intensity forecasts from the electricityMap API. These data are stored in a Data Service DB and versioned for access by the scheduling logic. When a developer commits new code, a CI trigger (e.g., from GitHub) is fired, orchestrated via Maestro. This launch triggers another bot that invokes a machine learning model, developed on Python and trained to predict hourly CO₂ emissions. Based on this forecast, the Agent identifies the optimal low-emission time window and schedules execution on the UiPath Test Suite. The scheduling thresholds and logic are managed via Data Service for maximum flexibility. -
Dossier Generation Agent
Once the tests are finished, a second agent is activated. It pulls screenshots, logs, and execution metadata from Storage Buckets. Using a Word-based reporting template, also stored in Storage Bucket, it dynamically generates a dossier that includes timestamps, results, execution conditions, and carbon footprint information. This report is exported as a PDF for compliance evidence and sent via mail.
Demo Video
Expected impact of this automation
- Time & carbon saved : carbon-aware scheduling + relevant-test re-execution cut up to 70 % CO₂ and ≈40 % runtime/cost per build
- Manual effort slashed : automated evidence capture trims QA audit work by 60-90 %, ending “audit-fatigue” cycles
- Dynamic & reusable : swap or add policy docs and the same RAG agent instantly produces new compliance + CO₂ reports—no retraining required
- Audit-grade accuracy: engine clusters failures and root-causes with high precision, feeding one-click audit reports
- Low-code, enterprise scaling: built in UiPath Agent Builder—agents orchestrate AI and RPA across any department or test domain.
UiPath products used (select up to 4 items)
UiPath Agent Builder
UiPath Automation Cloud™
UiPath Data Service
UiPath Test Suite
Automation Applications
Test Suite UiPath, potentially every applications
Integration with external technologies
Open AI, Python, Word
Agentic solution architecture (file size up to 4 MB)
Sample inputs and outputs for solution execution
input:
- Historical data of CO2 from Data Service
- Change Request on GitHub
- Template Dossier
Output
- Compiled Dossier