Intelligence Infrastructure for Africa's Energy Grid
Asoba builds the data and AI infrastructure that makes distributed renewable energy readable, tradeable, and insurable at scale. We unify fragmented OEM data into open standards, layer commercial intelligence on top, and give operators, regulators, and investors the clarity to act.
The Problem
The Renewable Grid-Edge is Choking on a Data Interoperability Crisis
Distributed renewable energy (DRE) has scaled from isolated rooftop systems to massive, multi-gigawatt installations. In South Africa alone, behind-the-meter rooftop solar has surged to 8,294 MW—now outstripping the country's entire grid-tied utility-scale renewable fleet. At the same time, NERSA has registered over 19.3 GW of private generation.
However, the digital operating model remains broken. The transition from a centralized monopoly grid to dynamic, peer-to-peer markets has triggered a severe grid-edge translation deficit:
- Inverter API Lock-In: Every major hardware manufacturer gates its telemetry behind proprietary, inconsistent formats and rate-limited cloud interfaces, forcing engineering teams into continuous "parser hell."
- Spreadsheet Chaos: Field dispatch notes, billing, and maintenance histories remain trapped in manual, unstructured formats, forcing operators to react to faults rather than proactively optimizing performance.
- Settlement & Trading Gaps: As competitive wholesale trading and private wheeling markets take flight under SAWEM, reconciling raw, complex half-hour utility meter files (like Eskom AMR profiles) with actual generation is settled manually via error-prone spreadsheets.
This data fragmentation introduces massive systemic risk, contributing to over $10 billion in global solar underperformance losses annually.
The Intervention
Three Layers of Intelligence Infrastructure
Open Protocol
ODSE
The open standard for energy asset data interchange. Apache-licensed, it standardizes inverter telemetry across manufacturers into a single schema. Install with pip install odse and start normalizing data from any OEM in minutes.
Commercial SaaS
Ona
The commercial renewable asset intelligence platform built on ODSE. Portfolio monitoring, AI-driven fault detection, day-ahead forecasting, yield optimization, and automated regulatory reporting — delivered as a multi-tenant SaaS dashboard.
AI Research Layer
Nehanda & Zorora
Zorora leverages our custom fine-tune of Qwen3.6 27B built on extra layers of SFT and DPO training for epistemic soundness critical for deep market research into emerging economy energy markets via satellite imaging, regulatory intelligence, and synthesis of hard-to-find financial data.
Our Strategy
Platform Formation and Research
Asoba’s strategy is built on a conviction that the energy transition in emerging markets will not be won by selling software licences — it will be won by whoever controls the data layer. We build open infrastructure first, then layer proprietary intelligence on top.
ODSE is free, Apache-licensed, and designed to eliminate the integration tax that every engineering team currently pays to wrangle incompatible OEM data formats. When the cost of standardization reaches zero, the barriers to trading, settlement, and AI-driven dispatch fall with it. Every megawatt normalized on ODSE expands the commercial network that Ona and Zorora operate within.
Nehanda is a 27B model fine-tuned across five stages of SFT and DPO specifically for RAG synthesis in high-stakes domains. It corrects false premises, preserves conflicting evidence, and cites sources inline. That epistemic discipline is what makes Zorora’s intelligence reports defensible to institutional clients who cannot afford to act on hallucinated data.
Operations
Cape Town HQ, Global Focus
Asoba is headquartered in Cape Town, South Africa, with operations across the SAPP region including South Africa and Zimbabwe, as well as in the United States. Our partnerships span energy operators, sustainability organizations, and technology providers who share our conviction that distributed energy needs an intelligence layer.