Energy and Utilities Reference Model | Capstera

The Energy and Utilities Business Architecture Reference Model brings the capabilities, value streams and data of an oil and gas company, a power generator or a utility into one package with the AI layer built in. It covers the work two or more of them share and three segments, Oil and Gas, Power Generation and Energy Supply, and Utility Networks, with 25 value streams, a 154-entity data model, 70 agent specification starters, 92 use cases and 61 controls, and every segment item is tagged so you can filter to your own. Business and enterprise architects load it into the Capstera platform with the import files, and their AI assistants query it through MCP.

What the Energy and Utilities Business Architecture Reference Model Is

This edition is the full business architecture of an energy company or a utility, with the AI layer built in. It includes everything in the capabilities map edition and builds the rest of the architecture around it. Its 25 value streams follow the work end to end, from an investment idea to a commissioned asset, an asset condition signal to a completed repair, a permit request to a closed permit to work, a process safety event to a closed action, a trade idea to a settled trade and a meter read to a paid bill. Each of the 125 stages is described as it works today and as it would work AI-native, with the human checkpoint stated at every stage, and 54 standard value streams come with them.

Under the value streams sits a data model of 20 subject areas, 154 entities, 1,051 attributes and 306 relationships, from capital projects, the asset register and planned outages to permits to work and isolations, control room operations, process safety, environmental operations, trading and settlement, meters and service points, and rates and billing, with each entity marked for sensitivity and for the data quality AI needs from it. The 70 agent specification starters state what each agent may read and do, its guardrails and when it hands back to a person, and they share one rule: agents never operate, start, stop, isolate or change the setpoints of plant, wells, pipelines or network equipment; never issue, extend or close a permit to work, override a safety system or release equipment to service; never decide a shutdown, a curtailment, a load-shedding or a supply interruption; never commit a trade, a nomination or a bid outside the approved limits; never change a meter reading, a bill or a customer's supply outside the approved rules, or decide a disconnection; never change an emissions, spill or water quality record, or decide what is reported about one; and never decide an action about a worker or a contractor. Protection systems that trip equipment, automatic generation control within limits the control room set, estimated reads under approved rules, billing runs from approved rates and limit checks in the trading system are run by the systems configured to them. The 92 use cases, seven of them practices the package records as not to be deployed, are written as business cases tied to capabilities, value streams, agents and data, and the 61 controls tie each risk tier to the evidence an owner keeps.

Each segment has value streams, entities, agents and use cases of its own. Oil and Gas brings five streams: a prospect to a drilled well, a well plan to a producing well, a reservoir review to a field production plan, a crude slate to a refined product, and a fuel order to delivered fuel. Power Generation and Energy Supply brings four: a generation plan to dispatched output, a renewable site to an operating plant, a supply offer to a supplied customer, and a demand response event to a settled payment. Utility Networks brings five: a connection request to an energized connection, a supply interruption to restored supply, a network plan to a built reinforcement, a distributed energy request to a connection, and raw water to delivered water. Their entities, such as wells, reservoirs and refinery run plans, generating units and storage assets, or feeders, substations and water quality samples, carry the same segment tag in the workbook.

The Problem It Solves

An energy company or a utility seldom lacks AI proposals. What it lacks is a way to see how one proposal touches the rest of the business: which stage of which value stream it changes, which data it depends on, who approves its output and which control shows it behaved as intended. The stakes are concrete, because a wrong output can let a job start under the wrong isolation, keep a safety system overridden past its end time, commit a trade outside its limits, send a household the wrong bill or misstate what a site released, and each of those decisions belongs to a named role.

The reference model connects those pieces before a project starts. A proposal maps to capabilities and stages, the stages name their data, the agents and use cases state their oversight and risk tier, and the controls and the operating model say who owns the result. Leadership can compare proposals from operations, engineering, safety, trading and customer teams on the same terms, and plan the shared work, such as maintenance, permits to work, planned outages and billing, once for every site and network instead of once per business unit.

What's Inside

Alongside the full capabilities map edition, the reference model holds the rest of the architecture:

• Everything in the Business Capabilities Map: 9 L1, 44 L2 and 176 L3 energy and utility capabilities, with definitions to L3 and 191 KPIs at L2, an AI profile for every L2, AI ratings for every L3, prioritization, the AI heatmap and a 40-question readiness assessment
• 25 energy and utility value streams, 125 stages with their steps, each described today and AI-native, with the human checkpoint at every stage, plus 54 standard value streams
• An AI-ready business data model: 20 subject areas, 154 entities, 1051 attributes and 306 relationships, with sensitivity and data-quality requirements
• 70 agent specification starters with tool permissions, guardrails and escalation paths
• 92 use cases as structured business cases
• 61 controls and responsible AI principles, tied to risk tiers
• An operating model: governance bodies, a RACI, an agent lifecycle, role impacts and new roles
• A three-wave roadmap
• A Welcome document, a leadership presentation template to adapt, the capability map deck, standalone documents for the capability definitions, KPIs, data model, agent specifications, use cases and AI governance, a value streams deck, the Excel workbook, the implementation guide, and import files for the Capstera platform
• Bonus files: an energy and utilities glossary, energy and utilities roles and how AI changes them, operating model archetypes for energy and utilities, an energy and utilities transformation overview and framework, A Practical Guide to Business Architecture, a capability modeling overview, an AI-ready capability profile template, capability relationship mapping templates, a framework-to-enablement deck, a capabilities to microservices and AI agents mapping example, and a business architecture leader role guide
• Free corrections, new versions free for 180 days, and MCP access for your AI assistants for the same 180 days

Every file names your organization and carries your license ID. Ratings describe a typical large enterprise and are a starting point for your own.

Use Cases

Designing agents for permit, safety and outage work. Start from an agent specification such as the permit to work preparation assistant, the barrier health monitoring agent or the restoration estimate assistant, and adapt its permissions, guardrails and escalation paths to your systems. Operations, process safety and network leaders then approve one document before anything is built.

Redesigning a value stream for AI. Take a stream such as a permit request to a closed permit to work, compare each stage today with its AI-native version, and decide where the human checkpoint stays. Each stage names who approves its outcome, so the redesign starts from a stated owner rather than an assumed one.

Building a use case portfolio. Use the 92 use cases as a starting portfolio, score them with your own value and data readiness ratings, and keep those that fit your segments. Each one already names the capabilities, value streams, agents and data it touches, from investment cases and inspection findings to meter data validation and leak surveys.

Setting up AI governance for safety, market and customer decisions. Adapt the responsible AI principles, the 61 controls, and the operating model's governance bodies and RACI to the way the organization already decides about plant, permits to work, trades, bills and disconnections. The risk tiers and the decision rights register tell the governance group which agents and use cases need its review and who approves each one.

Planning the data foundation. Compare the data model's entities, sensitivity markings and data quality requirements with the asset, condition, work, trading, meter and customer data you hold. The gaps become the data work that has to come before the use cases that depend on it, starting with asset records that two systems describe differently.

Briefing the executive team. Adapt the leadership presentation template to show the ranked capabilities, the first wave of use cases and the governance proposal. Leaders see the plan in terms of their own sites, networks, markets and decisions.

Who Uses It

Enterprise and business architects. Use the reference model as the architecture for the energy or utility business, link it to the enterprise model, and tailor the segments to the organization.

Chief operating officers, chief engineers and heads of operations. Use the asset, work and permit to work value streams and the roadmap to decide where AI changes maintenance, planned outages and field work first, and confirm that no agent operates equipment or issues a permit to work.

Heads of process safety and environment. Review the process safety event, emissions and leak repair streams and agents, decide which controls their teams will own, and confirm that no agent changes an emissions, spill or water quality record.

Heads of energy trading and commodity risk. Check the trade, nomination and settlement streams and the limit and credit monitoring agents, and confirm that every trade outside the approved limits goes to the person named for it.

Heads of metering, billing, customer operations and network operations. Use their streams and agents to plan AI in meter data, billing exceptions, collections, connections and supply restoration, with each disconnection and each bill corrected outside the rules decided by a person.

AI and transformation program leads. Use the use case portfolio, the operating model and the three-wave roadmap to staff and sequence the program across business units and segments.

Build It, Hire It, or Start Here

There are three routes by which an energy company or a utility reaches a business architecture with an AI layer, and they differ in how much of the effort goes into structure that every producer, generator and utility shares.

• Building it from a blank slate. The team elicits capabilities, maps value streams with each business unit, models the data, and only then designs agents, use cases and controls on top. Each layer depends on the one before it, so the AI work waits until the architecture is settled, and keeping the layers consistent takes repeated review.
• Hiring consultants to build it. A firm runs the elicitation and delivers the layers, often from its own internal starting point. Much of the effort pays for drafting what is common across the industry, and the links between layers are only as consistent as the firm's own checks.
• Starting from this model. The team starts with every layer already linked: capabilities to stages, stages to data, agents and use cases to both, and controls to risk tiers. Workshops go to setting aside the segments the organization does not run, adjusting ratings and oversight, and naming the approver for each decision.

Tailoring to your sites, assets and markets, and running the change, still needs people who know the organization, from inside it or outside. Starting here lets that work begin at the first meeting.

How to Get Started

Six steps take a team from the download to a first wave of AI work that leadership has approved:

1. Read the Welcome document and the implementation guide for the order in which the files are meant to be used.
2. Filter the workbook's Segment column to the shared items and the segments your organization runs.
3. Review the value streams with process owners and mark where your stages differ from the reference.
4. Check the data model against the asset, work, trading, meter and customer data you hold and list the gaps.
5. Score the use cases and agents with your own ratings, and name the approver for each decision before choosing the first wave.
6. Adapt the operating model, the controls and the leadership presentation, and take the plan to leadership.

Read the Public Sample

The sample follows one thread through the package, energy asset investment planning, and is 0.57% of the content. From the Energy Asset Investment Planning (EG-1.1) AI profile:

• Work patterns: Analysis & insight, Judgment-based decision, Content generation
• Target AI mode: Augment
• Human oversight: Human-in-the-loop
• AI fit / Value / Feasibility / Data readiness: 4 / 4 / 3 / 2 (1–5)
• Risk tier: Moderate
• Roadmap wave: Wave 1
• AI opportunity: Gather investment needs from condition assessments, failure history, integrity threats and demand forecasts into one list, each with its source and the risk if nothing is done
• What stays human: The Chief Financial Officer (enterprise) approves each capital investment after hearing the Head of Asset Management, prepared by the Head of Capital Projects; the Capital Plan Analysis Assistant ranks and drafts and never records a decision

The preview above shows more of the profile, the AI heatmap, an agent card and a value stream stage redesigned for AI. Any MCP client can also query the sample at https://platform.capstera.com/api/mcp/blueprints?package=eg without a key.

Connect Your AI Assistant

Your license includes access to the package through MCP (Model Context Protocol). Connect an MCP-capable AI assistant or your own agents, and they look up your licensed content as they work: one capability, agent or control at a time, scoped to your edition and stamped with your license ID. The key arrives with your order confirmation, and access runs for the 180 days in which new versions are free. Setup and limits: AI-ready license and MCP access.

Upgrade Credit

Bought the Business Capabilities Map in the last 90 days? The full price you paid counts toward this Blueprint. Email support@capstera.com with your order number for a one-time checkout code.

Frequently Asked Questions

What is the difference between the two editions?

The Business Capabilities Map covers the capability map, an AI profile for every L2 capability, prioritization, the heatmap and the readiness assessment. The AI Transformation Blueprint adds value streams redesigned for AI, the business data model, agent specifications, use cases, responsible AI principles and controls, the operating model and the roadmap, a leadership presentation template, and bonus files on roles, operating model archetypes and transformation.

Can I upgrade from the Business Capabilities Map to the Blueprint later?

Yes. The full price you paid for the Business Capabilities Map counts toward the Blueprint if you upgrade within 90 days. Email support@capstera.com with your order number for a one-time checkout code.

What happens after 180 days?

Your license to use the files does not end. Corrections are free for as long as you use the files. New versions released within 180 days of purchase are free, and MCP access runs for the same 180 days. There is no renewal after that: you can buy a new version at 25% off its price. Orders placed before October 27, 2026 keep the 12 months they were sold with.

Can I load the content into our own AI tools?

Yes, within the license. The Enterprise license allows internal tools, including your own assistants and agents. Neither license allows training, fine-tuning or prompting AI systems for third parties.

Does the package cover AI laws and regulations?

No, by design. It is business architecture: capabilities, value streams, data and the AI layer describe what an energy company or a utility does and how AI fits, and they hold whichever laws apply where you operate. The risk tiers, principles and controls are durable practices your compliance team maps to your own jurisdictions.

Why not just ask ChatGPT or another AI assistant?

An AI assistant can sketch value streams or suggest use cases for an energy company or a utility, and that is useful for a first look. A reference model has to hold together across its layers. Here every link between capabilities, value stream stages, data entities, agents, use cases and controls is checked before release, IDs stay stable, every agent and use case carries a risk tier, and segment items are tagged so they can be filtered instead of guessed. A generated answer changes with each prompt and checks none of those links, which matters once operations, process safety, trading and finance leaders have to approve the plan.