Level 0 – 5: Challenges, Opportunities & The P2AgentX Platform (Adapted from SAE Automation Framework)
Executive Summary
The global solar energy sector is scaling rapidly, with gigawatt-scale farms increasingly common across Australia, the United States, Europe, and Asia. As these assets grow in size and complexity, traditional manual operations become unsustainable — not only in cost, but in safety, speed of fault response, and long-term energy yield optimisation.
This document presents the five-level Autonomy Framework for solar farm operations, adapted from the SAE (Society of Automotive Engineers) classification system widely used in autonomous vehicle development. Each level defines a progressively higher degree of machine-led decision-making, reducing reliance on human intervention while increasing operational intelligence.
For each level, this document outlines:
- The operational characteristics and scope of automation
- The key challenges faced by operators at that level
- The opportunities that automation unlocks
- The enabling features and technology capabilities required
The second section of this document details how the P2AgentX platform — specifically P2Chat and P2Dingo — directly addresses the challenges at each level and serves as the accelerant for operators moving up the autonomy ladder.
This stepwise approach allows the industry to build trust in automation while progressively de-risking technology deployments.
The Five levels of Solar Farm Autonomy
The table below provides a high-level overview of all five autonomy levels. Detailed analysis of each level — including challenges, opportunities and key enablers — follows in subsequent sections.
| Level | SAE Classification | Description | Human Role |
| 0 | No Automation | All monitoring, inspections and reporting performed manually, supported only by basic SCADA dashboards. | Fully responsible for all decisions and actions |
| 1 | Assisted Operation | Tools gather data and display KPIs; humans interpret results and act. No automated actions. | Reviews data; makes all decisions |
| 2 | Partial Automation | AI detects faults, generates reports and job cards. Some limited automation. | Approves and executes AI suggestions |
| 3 | Conditional Automation | System executes diagnostics and dispatches actions under known conditions. | Monitors and handles exceptions |
| 4 | High Automation | Full AI + robotics perform all routine monitoring, inspection and planned maintenance. | Strategic oversight and escalations |
| 5 | Full Automation | End-to-end autonomous operation including adaptive strategies, unstructured inspection and corrective maintenance. | Minimal to none; governance only |
Level 0 – No Automation
| Level 0 No Automation All monitoring, inspections, reporting and repairs are carried out manually by humans, supported only by basic SCADA (Supervisory Control and Data Acquisition) dashboards. No automated actions or AI assistance. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| Level 0 | • Slow fault detection — faults may go unnoticed for days • High labour costs and operator fatigue across large sites • Inconsistent inspection quality and reporting standards • No centralised visibility across distributed assets • Inability to scale to multi-site portfolios • Safety risks from manual physical inspection | • Digitisation of any manual process offers immediate ROI • Even basic sensor data collection establishes a baseline for future AI • Strong opportunity to define and standardise workflows before automation • Training data generation for future ML models | • Manual log books and spreadsheets • Site walk inspections and ad hoc fault recording • Phone-based escalation chains • Paper-based or PDF job card management |
Level 1 – Assisted Operation
| Level 1 Assisted Operation Tools gather data and display Key Performance Indicators (KPIs) via SCADA systems and sensor dashboards, but humans interpret all results and act. No automated actions are taken. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| Level 1 | • Data overload — operators flooded with alerts and metrics without intelligent prioritisation • Siloed systems with no unified interface for cross-asset analysis • Alert fatigue leading to missed critical events • Limited ability to identify patterns or predict failures • High cognitive load on O&M engineers and shift supervisors • Reactive rather than proactive maintenance posture | • First step toward data-driven decision-making • Historical data collection builds foundation for predictive analytics • Improved documentation and compliance reporting • Faster fault identification compared to fully manual operations • Better root cause analysis capability using trend data | • SCADA and DCS integration for real-time monitoring • Centralised dashboards and KPI visualisation • Basic alarm management and notification routing • Historical data storage and trend analysis • Remote access and monitoring capabilities |
Level 2 – Partial Automation
| Level 2 Partial Automation AI identifies faults, suggests corrective actions, and generates job cards for humans to review and execute. Some limited automated responses are possible under controlled conditions. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| Level 2 | • Change management — operators must trust and adopt AI recommendations • Accuracy of AI fault classification requires validated training data • Integration complexity with legacy SCADA and asset management systems • Workflow design to handle AI-generated job cards alongside manual tasks • Determining which automation actions are safe without human review • Compliance and audit trail requirements for AI-initiated actions | • Significant reduction in mean-time-to-detect (MTTD) for faults • Automated job card generation reduces administrative burden by 40–60% • AI pattern recognition identifies systemic issues invisible to human operators • Freed operator time can be redirected to strategic and complex tasks • Improved SLA compliance through faster fault routing • Foundation for predictive maintenance programmes | • AI-powered fault detection and classification • Automated job card and work order generation • Natural language interface for O&M queries • Integration with CMMS and asset management platforms • Anomaly detection with confidence scoring • Automated report generation and stakeholder communications |
Level 3 – Conditional Automation
| Level 3 Conditional Automation The system autonomously executes diagnostics and dispatches actions under known, pre-defined conditions. Human oversight is required for exceptions and novel situations. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| Level 3 | • Defining safe operating envelopes for autonomous action with precision • Managing liability and responsibility when automation takes action • Ensuring fail-safe escalation paths when conditions fall outside parameters • Cybersecurity risks increase as systems gain authority to act • Operator de-skilling over time as routine tasks are automated • System reliability requirements are significantly higher than at lower levels | • Dramatic reduction in response times for known fault types • 24/7 automated monitoring without shift-dependency • Significant labour cost reduction for routine O&M • Consistent execution quality — no human variability • Scalability to larger portfolios without proportional headcount growth • Performance data generated for continuous model improvement | • Autonomous dispatch of maintenance crews for known fault types • Conditional logic engines with configurable rule sets • Bi-directional integration with field management systems • Real-time human-machine collaboration interfaces • Automated exception escalation and notification workflows • Audit trails and explainability for all automated decisions |
Level 4 – High Automation
| Level 4 High Automation Full AI and robotic systems perform all routine monitoring, inspection and planned maintenance with minimal human input. Humans are responsible for strategic decisions and rare escalations. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| SAE Level 4 | • High capital investment in robotic and autonomous inspection hardware • Complex multi-system orchestration across AI, robotics, and SCADA • Regulatory frameworks for autonomous physical operations are still maturing • Extensive validation and testing required before deployment • Edge cases and unstructured fault scenarios remain human-dependent • Workforce transition — significant retraining required for remaining operators | • Near-reduction of routine labour costs at scale • Continuous inspection cycle (not periodic) improves fault detection rates • Higher energy yield through proactive maintenance and soiling correction • Operator role shifts to high-value strategic and engineering work • Data richness enables next-generation predictive and prescriptive analytics • Competitive differentiator for asset owners and O&M contractors | • Autonomous drone and ground-robot inspection fleets • AI-driven thermal, soiling, and structural analysis • Automated work scheduling and resource optimisation • Predictive maintenance with lead-time scheduling • Digital twin integration for simulation and planning • Integrated work management and analytics dashboards |
Level 5 – Full Automation
| Level 5 Full Automation End-to-end autonomous operation across the full asset lifecycle. The system handles adaptive strategies, unstructured inspection scenarios and complex corrective maintenance with minimal to no human involvement. | |||
| ⚠ Challenges | ✦ Opportunities | ★ Key Features | |
| Level 5 | • Achieving sufficient AI reliability across all edge cases and novel fault types • Plant equipment are not designed for robots to handle, and may require redesign Regulatory and insurance frameworks may not yet support fully autonomous operation • Significant upfront investment in both software and physical infrastructure • Public and investor confidence in autonomous system safety and governance • Defining meaningful human governance roles in a near-zero-intervention model • Ensuring system resilience against cyber threats, environmental events and hardware failures | • Fully scalable O&M model — portfolio growth without corresponding cost growth • Highest possible energy yield through continuous optimisation • New commercial models — performance-based O&M contracts become viable • Operators become technology companies rather than labour-intensive service firms • Unlocks autonomous grid services and energy market participation • Industry leadership and first-mover advantage in a defining sector transition | • Autonomous command and control centre with AI orchestration • Self-learning models that improve with every operational cycle • Full digital twin with physics-based simulation • Autonomous regulatory reporting and compliance management • Adaptive energy yield optimisation algorithms • End-to-end lifecycle management from commissioning to decommissioning |
The P2AgentX Platform – P2Chat & P2Dingo
P2AgentX has built a suite of AI-powered products specifically designed to accelerate the journey from Level 0 through to Level 5 autonomy in solar farm operations. The two flagship products — P2Chat and P2Dingo — are deeply integrated, complementary tools that address the full spectrum of autonomy challenges identified in above section.
Together, they form a unified intelligent operations platform: P2Chat serves as the conversational intelligence and orchestration layer, while P2Dingo provides the autonomous sensing, inspection and physical operations capability.
P2Chat — Conversational AI Operations Interface
An AI-powered conversational interface that connects operators to SCADA data, historical trends, fault analysis and workflow automation through natural language.
| Automation Levels | Core Capabilities | Business Benefits |
| Primarily Levels 1–5 (with early deployment value at Level 0) | • Natural language querying of SCADA, historian and asset data • Automated fault identification, classification and prioritisation • AI-generated job cards and work order creation • Orchestration of O&M workflows across human and autonomous agents • Integration with existing CMMS, ERP and document management systems • Automated reporting, stakeholder communications and audit trails • Predictive analytics and trend-based maintenance recommendations • Multi-site visibility and portfolio-level performance monitoring | • Eliminates alert fatigue through intelligent prioritisation • Reduces administrative burden by up to 60% through automated documentation • Accelerates fault detection and response, improving energy yield • Scalable to any portfolio size without proportional headcount growth • Reduces dependency on specialist knowledge held by key individuals • Provides explainable, auditable AI decisions for regulatory compliance • Measurable ROI from day one of deployment |
P2Dingo — Autonomous Inspection & Field Operations
An autonomous inspection and field operations platform that deploys AI-guided thermal scanning, drone inspection and robotic patrols to deliver continuous physical asset monitoring.
| Automation Levels | Core Capabilities | Business Benefits |
| Primarily Levels 2–5 (v1: Semi-auto; v2: Conditional; v3: Full) | • AI-guided thermal and visual inspection via drone and ground-based robotics • Automated anomaly detection for soiling, degradation and electrical faults • Autonomous patrol scheduling and execution under AI direction • Real-time integration with P2Chat for seamless digital-physical workflow • v1: Semi-autonomous inspection with AI-flagged findings for human review • v2: Conditional autonomous patrols with automated anomaly identification • v3: Fully autonomous field operations with minimal human oversight • Structured data output feeding continuous model training and improvement | • Continuous inspection cycle replaces periodic manual surveys • Early fault detection through thermal imaging reduces losses from degraded panels • Consistent inspection quality eliminates human variability • Significant reduction in physical inspection labour costs • Safer operations by reducing human exposure to electrical and environmental hazards • Rich data feeds improve predictive maintenance models over time • Scales to utility-scale sites without proportional inspection cost growth |
How P2Chat and P2Dingo address each autonomy level
| Level | Classification | P2Chat Role | P2Dingo Role |
| 0 | No Automation | Digitises manual logs; provides first structured data interface | Not yet deployed — baseline data collection underway |
| 1 | Assisted Operation | Unifies SCADA, historian and asset data via conversational AI; eliminates alert fatigue | Not yet deployed — SCADA integration feeds P2Chat |
| 2 | Partial Automation | Identifies faults, generates job cards, suggests corrective actions; integrates with CMMS | Not yet deployed – Independent inspection |
| 3 | Conditional Automation | Orchestrates O&M workflows; coordinates human-machine handoffs; manages dispatch | ONE: Conditional patrols execute autonomously; findings trigger P2Chat workflows |
| 4 | High Automation | Supervises operations; integrates analytics; manages escalation and stakeholder reporting | TWO: Full autonomous inspection; predictive maintenance scheduling |
| 5 | Full Automation | Autonomous command centre; AI-led decision-making across portfolio; strategic reporting | THREE: Fully integrated physical autonomous operations with adaptive AI field direction |
Conclusion
The transition from manual solar farm operations to fully autonomous management is not a distant aspiration — it is an active and accelerating commercial reality. Operators who delay this transition risk increasing operational costs, lower energy yields, and competitive disadvantage against peers who are already deploying AI-powered O&M solutions.
The five-level framework presented in this document provides a clear, structured pathway for operators at every stage of maturity. Whether an organisation is running entirely manual operations today or looking to close the gap to full Level 5 autonomy, each step offers tangible, measurable business value.
P2Chat and P2Dingo are built specifically for this journey. P2Chat delivers intelligent, conversational operations management that connects every data source, workflow and stakeholder through a single AI interface. P2Dingo delivers the physical autonomous inspection and field operations capability that closes the loop between digital intelligence and on-the-ground action.
Together, they represent the most complete AI-powered solution for solar farm operators today — designed not just to automate what exists, but to reimagine what solar operations can become.
