Status: Reference architectures and demonstration designs — not production deployments.
Utilities make the agentic-enterprise control problem concrete. Grid operations, field service, customer programs, meter telemetry, and compliance combine delegated authority with safety, regulatory, and operational consequences. This portfolio demonstrates the integration and governance architecture required to introduce AI agents into that environment responsibly.
It is a set of reference architectures and demonstration implementations — MuleSoft integration, API and event designs, and governed agent patterns — spanning five utility domains. Every showcased capability links to an inspectable artifact, with concepts, modeled scenarios and implementation gaps labeled explicitly. Where an artifact is a design target or a modeled scenario rather than a measured result, it is labeled as such.
These are personal reference architectures reflecting independent professional research and experience. They do not represent product commitments, customer implementations, or the views of any employer. Agentforce and MuleSoft are described factually as platform technologies; no vendor endorsement is implied.
| Domain | What it covers | Repositories |
|---|---|---|
| Grid intelligence & resilience | Storm/outage response, grid data integration, agent-assisted operations | utility-ai-mulesoft-api, energy-api-evolution |
| Field operations & assets | Crew dispatch, field agents, vegetation management, work & asset management | field-operations-support-agent, vegetation-management-process-api, Work-and-Asset-Management-API |
| Customer & energy programs | Customer service, energy-efficiency enrollment | utility-customer-service-process-api, Energy-Efficiency-Program-Enrollment-API |
| Smart meters & telemetry | AMI readings, events, tampering/outage signals, data simulation | SmartMeterAIAPI, SmartMeterAsyncAPI, Utilities-Generator-API — see note¹ |
| Compliance & critical infrastructure | Regulatory process automation, audit evidence | Utilities-Compliance-Management, trusted-context-level-4-digital-labor |
¹ Earlier smart-meter experiments are planned for consolidation into a unified reference
architecture and are not included in the showcased portfolio. The showcased smart-meter
components above (SmartMeterAIAPI, SmartMeterAsyncAPI, Utilities-Generator-API) remain
the inspectable artifacts. See docs/repository-inventory.md.
Each repository states its business problem, target user, implementation status, what is
actually built, a demonstration path, and its governance boundaries. Current status and
disposition for every repository live in
docs/repository-inventory.md.
Scope note: this is an electric-utility portfolio. Oil & Gas material is not part of it and is reserved for a separate, future Oil & Gas hub; it is not showcased here. The inventory lists only the showcased electric-utility repositories.
- Integration is the foundation. MuleSoft and API/event architecture are how utility systems of record, field systems, meters, and customer channels are safely connected.
- Agents are governed, not autonomous-by-default. Where AI agents act, they act inside explicit authority, audit, and safety boundaries — the "governance overlay," not uncontrolled automation.
- Reference vs. production is always distinguished. Nothing here claims a real deployment, a compliance certification, or a measured savings figure unless a reproducible artifact backs it.
Four kinds of asset appear across the portfolio, and each repo says which it is: domain integration assets · agent-enabled use cases · governance overlays · EAA-aligned reference architectures.
This portfolio holds itself to a single rule: a claim is either backed by an inspectable artifact, labeled as a modeled scenario / design target, or removed.
- No "production-ready," "compliant," or "certified" without reproducible evidence.
- No savings, dollar-value, accuracy, uptime, latency, or throughput numbers presented as measured results unless the measurement is reproducible and linked.
- No fictitious customers, deployments, domains, or support channels.
- Simulated data is labeled simulated.
The showcased component repositories have been reconciled to this standard — aspirational figures (inflated API-count headlines, dollar-value claims) have been removed or relabeled at the source. This hub describes each repository by what is actually in it.
See docs/evidence-standard.md.
docs/portfolio-guide.md— the guided tourdocs/capability-map.md— capabilities → repositoriesdocs/repository-inventory.md— every repo, status, dispositiondocs/architecture-principles.md— the integration + governance principlesdocs/evidence-standard.md— the claim rule in fulldocs/terminology.md— EAA / utility / MuleSoft termsdocs/roadmap.md— what is built vs. plannedarchitecture/— capability map + reference-architecture diagramsdemos/— storm response, field ops, customer service, efficiency, smart meter, compliance
Documentation and diagrams: CC BY 4.0 (see LICENSE-docs). Any reusable
code/specs in this hub: Apache-2.0 (see LICENSE). Individual repositories carry
their own licenses; see each repo. Several sibling repositories currently ship without an
SPDX license — see the inventory license-gap note.
Michael K. Saleme is an enterprise architect with 30 years of integration, data and automation experience across Energy, Utilities and Consumer Goods. This portfolio applies Enterprise Agent Architecture to electric-utility operations.
- Portfolio index: https://github.com/msaleme/start-here
- Enterprise Agent Architecture: https://cognitivethoughtengine.com/eaa
- LinkedIn: https://www.linkedin.com/in/mikesaleme/