Asoba Asoba
Commercial Optimization

Optimize your assets, trading, and ancillary services.

Independent power producers lose margin when forecasts are generic, underperformance goes unattributed, and settlement runs on spreadsheets. Ona closes the gap between physical generation and financial outcomes — with site-tuned intelligence you can use in the dashboard or embed via SDK.

Site-tuned forecasting from day one Per-inverter revenue attribution UI + SDK access
🧠 PV Insight Assistant (JEPA + RAG) READY Ask PV Insight for troubleshooting or BOM advice... Ask ⚡ Analyze Anomaly 🔧 Spare Parts 📖 Manual Steps DIAGNOSTIC FINDING (JEPA TELEMETRY MATCH) Inverter INV-04 String 2 DC Voltage Shortfall CITED OEM MANUAL SOURCE: Huawei SUN2000-330KTL Manual §4.2 (Table 4-1 Troubleshooting) NEHANDA RECOMMENDED BOM REPLACEMENTS: • 1x DC Fuse 1500V 30A (SKU #HW-8812) — $45.00 • 1x MPPT Input Board v2 (SKU #HW-9904) — $420.00 + Add Recommended Parts to Maintenance Plan

The Business Problem

Trading, wheeling, and ancillary service decisions depend on numbers your current stack can't produce reliably — or produces too late to act on.

Generic forecasts miss site reality

Industry models need 24+ months of local data for bankable projections. Without site-tuned forecasts grounded in your telemetry, trading bids and PPA settlements are built on assumptions — not your asset.

Underperformance hides in portfolio averages

When 4 of 7 inverters underperform but portfolio KPIs look acceptable, cumulative revenue loss goes unnoticed. You need per-inverter attribution with financial impact — not threshold alarms.

Settlement and dispatch need deterministic execution

Cloud lag costs money in trading intervals. Wheeling certificates, municipal reconciliation, and dispatch optimization require sub-fifteen-minute pipelines — with identical inputs producing identical control decisions every time.

Ona: Bounded Intelligence for Asset Optimization

A JEPA + LLM hybrid architecture — System 1 runs continuously on every inverter; System 2 triggers only when severity earns the compute. Both ship as composable SDK modules.

System 1

JEPA World Model — Predictive AI

Continuous physics-conditioned tracking on 5-minute telemetry. Site-tuned generation forecasts, per-inverter performance ratios, and financial impact quantification — the numbers that feed trading bids, wheeling settlement, and ancillary service dispatch.

Transfer learning from regional baseline datasets bypasses the standard two-year data-accumulation requirement. Bankable projections in weeks, not years.

System 2

Nehanda Reasoner — Retrieval-Grounded Synthesis

When JEPA flags severity, Nehanda produces cited diagnostics: OEM troubleshooting steps, tariff context, and regulatory interpretation — bounded to retrieved evidence, not free-text chat. The anomaly writes its own query; Nehanda synthesizes over manufacturer manuals and policy corpora.

Decide

Deterministic Decision Layer

Dispatch optimization, trading interval bids, and wheeling certificate generation run on deterministic solvers — identical inputs yield identical outputs. Generative models occupy Orient only; they never hold closed-loop actuation authority over physical assets.

Read the Bounded Intelligence architecture →

Intraday Trading & Settlement

From day-ahead bidding to automated wheeling reconciliation

Unpredictable solar yield leads to costly market imbalance fines and disputed off-taker invoices. Ona connects physical generation directly to financial accounting—feeding site-tuned yield curves into 15-minute trading bids and generating audit-ready settlement certificates automatically.

  • Imbalance Penalty Avoidance: High-precision 15-minute intraday yield forecasts protect margins during sudden cloud-cover ramps.
  • Automated Wheeling Certificates: Instant reconciliation of multi-party wheeling credits across municipal and utility grid nodes.
  • Financial Variance Visibility: Track expected vs. realized revenue in real time with automatic PPA tariff indexing.
Ona forecasting and settlement intelligence dashboard

Real-World Revenue Recovery

See how Ona turns operational data into quantified financial value across forecasting, data reconstruction, anomaly detection, and edge deployment.

Gold Mine Power Production Forecasting

Reliable Yield Forecasting in 1 Week with Only 3 Months of Data — Enabling Bankable Revenue Projections

IndustryCommercial
Capacity2 MW
LocationGweru, Zimbabwe

Challenge

Only 3 months of historical data available for a 2 MW solar plant. Industry standard requires 24+ months for accurate forecasting models, making bankable revenue projections impossible with limited local data.

0.94
R² Accuracy
1 wk
Time to Bankable Forecast
Daily Power Production Profile Forecast vs Actual (Gold Mine 2 MW) 2000 1500 1000 500 0 Forecast Actual (R² = 0.94)

Lost Data Imputation

Overcoming 65% Data Loss: Preserving Revenue Reporting Through Infrastructure Failure

IndustryIndustrial
Portfolio SizeMulti-MW
Data Loss65% Missing

Challenge

Severe data loss — 65% missing data points across a multi-MW portfolio. Sensor and telemetry failures made traditional monitoring and revenue reporting impossible.

100%
Data Reconstruction
96%
Faster Fault Detection
Telemetry Imputation & Yield Recovery Reconstructing 65% Portfolio Sensor Blackout 2.0 1.5 1.0 0.5 0 06:00 09:00 12:00 15:00 18:00 65% SENSOR BLACKOUT Live Telemetry Sensor Outage Imputed Yield

Embedded Generation Revenue Recovery

Quantifying Hidden Revenue Loss: Per-Inverter Attribution Turns Underperformance Into Recoverable Dollars

IndustrySolar IPP
Capacity2.31 MW
Inverters7 × 330 kW

Challenge

System Performance Ratio at 74.1% vs 80% contractual target. 4 of 7 inverters underperforming with an energy shortfall of 291,279 kWh (32.2%). Without per-inverter attribution, the cumulative revenue loss went unnoticed by ownership.

$11k+
Annual Revenue Opportunity Identified
14,642
Anomalies Detected

Read the full case study →

Per-Inverter Performance Ratio INV-190 54.9% INV-191 68.2% INV-194 76.0% INV-193 82.4%

Access via UI and SDK

Operators use the Ona Platform dashboard. Engineers embed intelligence directly into trading systems, ERP workflows, and custom automations.

Ona Platform

Dashboard & Tools

Forecasts, monitoring dashboards, settlement views, and Energy-at-Risk reports — no code required.

  • Day-ahead forecasting via forecast.asoba.co
  • Per-meter anomaly monitoring & alerts
  • Portfolio performance & settlement reports
Ona SDK

Python & JavaScript SDKs

REST APIs and SDK clients for production integrations — compose forecasts, OODA alerts, and Nehanda diagnostics into your existing stack.

Pricing

Start with per-meter anomaly monitoring. Add carbon, PPA, and trading modules as your optimization stack grows. Available via UI and SDK.

Per-Meter Anomaly Monitoring

Ongoing Monitoring
$15 / meter / month

Continuous physics-based anomaly detection across your inverter fleet. Get alerted before a hardware fault becomes a month of lost production — with automatic financial impact tracking in your local currency.

  • 5-minute telemetry ingestion
  • Physics-based fault detection (JEPA world model)
  • Inverter state fault classification
  • Email & webhook alerts
  • Monthly delta performance report
  • Rolling financial impact tracking
+

Add-On Modules

Optimization Stack
From $50 / module / month

Extend monitoring with the modules your use case demands — trading, settlement, carbon, and PPA. Each add-on layers on existing telemetry with no re-integration.

  • Carbon Monitoring — Scope 2 emissions, ESG & investor reporting
  • PPA Management — Settlement-grade metering, offtaker billing
  • Trading & Settlement — Wheeling certificates, dispatch optimization, regulatory intelligence
Contact Sales
Built on ODSE — Free & Open Source

All monitoring and add-ons run on the Open Data Schema for Energy (CC-BY-SA / Apache 2.0). Install the standard for free. When you're ready for continuous intelligence, it scales up with zero integration cost.

Start optimizing your asset and trading decisions.

Connect via the Ona dashboard or embed intelligence through the SDK. We'll map your forecasting gaps and show you exactly what site-tuned optimization delivers.

Get Help & Stay Updated