AIO & STPIS Compliance

Annual Information Obligations (AIO)

Annual Information Obligations are the detailed data submissions that DNSPs must make to the AER each year. Unlike RINs (which are requested), AIOs are standing obligations defined in the Electricity Distribution Network Service Providers Service Target Performance Incentive Scheme.

AIO Categories

Energy Copilot tracks AIO obligations across 8 categories:

Category	Typical Data Points	Filing Deadline
Customer Service	Call wait times, complaints, connection times	31 October
Reliability Metrics	SAIDI, SAIFI, MAIFI by feeder	31 October
Asset Information	Asset registry, condition assessments	31 October
Financial Information	Actual capex and opex by category	31 October
Network Performance	Outage management data	31 October
DER Information	Connections, hosting capacity	31 October
Safety Data	Electrical incidents, near misses	31 October
Environmental Data	Vegetation management, bushfire risk	31 October

AIO Hub (`/dnsp/aio`)

The AIO Hub provides end-to-end AIO management:

Filing status tracker: each AIO item with completion status
Data validation: checks data against AER schema requirements before submission
Draft generator: Claude Sonnet 4.5 generates narrative sections from structured data
Submission pack: assembled AIO submission ready for AER portal upload

Service Target Performance Incentive Scheme (STPIS)

STPIS creates financial incentives for DNSPs to improve network reliability above their AER-set targets. Under-performing DNSPs face revenue reductions; over-performing ones earn bonuses.

STPIS Metrics

Three reliability metrics are tracked under STPIS:

Metric	Full Name	Definition	Direction
SAIDI	System Average Interruption Duration Index	Total customer-minutes interrupted / Total customers	Lower = better
SAIFI	System Average Interruption Frequency Index	Total customer interruptions / Total customers	Lower = better
MAIFI	Momentary Average Interruption Frequency Index	Momentary interruptions (<1 min) / Total customers	Lower = better

Revenue Impact

The STPIS incentive payment is calculated as:

STPIS_Revenue_Adjustment = Incentive_Rate × (Target - Actual) × Customer_Base

# Example (simplified):
# SAIDI target: 120 min/customer/year
# Actual SAIDI: 115 min/customer/year
# Incentive rate: $100,000 per customer-minute improvement
# Revenue bonus: $100,000 × (120 - 115) = $500,000

The maximum revenue adjustment (cap) is typically ±1% of the DNSP’s allowed annual revenue.

STPIS Dashboard (`/dnsp/stpis`)

Current year performance: SAIDI, SAIFI, MAIFI vs target — year-to-date
Revenue impact: estimated incentive payment based on current trajectory
Feeder ranking: which feeders have the worst reliability performance?
Outage categorisation: planned vs unplanned, SAIDI contribution by cause
Time series: 5-year historical trend

Screenshot: STPIS dashboard showing SAIDI YTD vs target gauge, feeder ranking table, and 5-year trend chart.

-- SAIDI performance by feeder, current year
SELECT
    feeder_id,
    feeder_name,
    zone_substation,
    saidi_ytd_minutes,
    saidi_target_minutes,
    saidi_variance_minutes,
    saifi_ytd,
    customer_count,
    estimated_revenue_impact_aud
FROM energy_copilot.gold.dnsp_stpis_metrics
WHERE financial_year = YEAR(CURRENT_DATE())
ORDER BY saidi_ytd_minutes DESC
LIMIT 20;

AI-Powered STPIS Anomaly Detection

The STPIS Anomaly Detection capability uses an Isolation Forest model combined with Z-score ensemble detection to identify anomalous reliability performance patterns across the 6-DNSP peer group.

What It Detects

Sudden SAIDI spikes: unusually high outage duration in a specific period
SAIFI anomalies: unexpected increase in interruption frequency
Peer outliers: a DNSP performing significantly worse than the peer group norm
Data quality anomalies: values that are likely reporting errors vs genuine performance issues

Detection Performance

Metric	Value
Detection rate	93.4%
False positive rate	4.7%
True negatives	95.3%
Z-score threshold	2.5σ
Isolation Forest contamination	0.05

See STPIS Anomaly Detection ML Model for technical details.

AIO Draft Generator

The AI Draft Generator uses Claude Sonnet 4.5 to generate draft narrative sections for AIO submissions. Given structured data (SAIDI numbers, asset counts, expenditure figures), it produces compliant, professional text in the AER’s preferred format.

See AIO Draft Generator (Claude) for full documentation.

API Endpoints

# AIO filing status
GET /api/dnsp/aio/status?year=2025&dnsp=ausgrid

# STPIS metrics
GET /api/dnsp/stpis/metrics?dnsp=ergon&year=2025

# Feeder ranking by SAIDI
GET /api/dnsp/stpis/feeders?dnsp=energex&year=2025&sort=saidi_desc

# Revenue impact estimate
GET /api/dnsp/stpis/revenue-impact?dnsp=essential_energy&year=2025

# STPIS anomaly detection results
GET /api/dnsp/stpis/anomalies?dnsp=ausnet&year=2025

# Generate AIO draft section
POST /api/dnsp/aio/generate-draft
{
  "dnsp": "sa_power_networks",
  "section": "reliability_performance",
  "year": 2025
}