AI at OG&E
Transforming grid operations — from the media wall to the switch order.
About OG&E
A quick tour of our service territory and Distribution Control Center.
- 95% of 542 substations are SCADA-controlled
- ~1,400 Intellirupters across the Distribution System
From the high side of substation transformers with low-side voltages of 34.5 kV and below.
Full OSI ADMS — including FLISR and SOM.
- Distribution Service Operators (Control 1) — single-circuit switching and dispatching of trouble tickets [Manage OMS]
- Distribution Operators (Control 2) — multi-circuit and substation switching [Manage DMS]
- Shift Supervisor — supervises real-time operations
- Grid Supervisors — direct supervision of operators
- Manager, Distribution Grid Operations
- Distribution Support Engineers
- DMS Engineers
- Analyst
- Human Performance / Training Coordinator
What We'll Cover
A quick preview of what's ahead.
The Platform
Meet Nighthawk — the in-house web platform that hosts every app in this deck.
Project 1 — The Media Wall
From a cluttered 4×2 grid to a glanceable, adaptive wall — redesigned and rebuilt with AI.
Project 2 — Resource Availability
An in-house dispatch app with measured ROI and a direct line to CMI / SAIDI.
Project 3 — Operator Switching Dashboard
Making switching quality measurable, peer-anchored, and safety-reinforcing.
Behind the Scenes
What AI did across these projects, our seven-step process, and the six-month journey.
Your Questions
What would translate to your DCC? Let's trade notes.
The Platform Behind Every Project: Nighthawk
OG&E's in-house web platform — built around how we work, with live data from every enterprise system we run.
Fits Our Workflow
Shaped to how we already work — no vendor software to retrain people on.
Uses Our Real Data
Live integrations with Oracle, SAP, ADMS, CAD, and any other source the problem needs.
Secure Single Sign-On
Integrated with Azure AD — one OG&E login, every Nighthawk app, no separate credentials.
AI Accelerates Build
Describe the need — AI helps design, code, and integrate.
Project 1 — The Media Wall
The Activu wall's job is to deliver a single, shared message to everyone in the DCC — Distribution Service Operators, Distribution Operators, Supervisors, and other leadership. It has to be readable from any seat in the room.
- Circuit Operations — dense table with no column headers and colors no one understood. Half the list scrolled out of view.
- Duplicate information taking up too much space.
- Five maps (temp, wind, rain, radar, outage) — each too small to read.
“Operators learned to ignore the wall. Leadership began to ask if we needed one at all.”
Three-section layout with a live Radar Map in the center. The map changes state visually to ensure situational awareness.
Project 2 — Resource Availability
- 75–105 emails/week processed manually
- 5–7 minutes per dispatch decision (up to 15 during complications)
- Conflicting answers across five different lists
- Single source of truth: no one's inbox
- No historical data — nothing captured to analyze later
- One unified application: on-call + leave calendars
- 30 seconds to identify and reach the right person
- Real-time CADS status integrated directly
- Single source of truth, always accurate
- Complete audit trail — every schedule, approval, and dispatch captured and queryable
Build vs. buy — the capital story
A COTS product would have shipped fast — and demanded that our operators change how they work to match it. We built it in-house instead, encoding the process we already ran. The numbers:
This isn't just a cost story — it's a reliability story. Every minute we shave off “who do we call?” flows directly into CMI (Customer Minutes of Interruption) and SAIDI (System Average Interruption Duration Index) — the two metrics utility leaders are held accountable to. When dispatch goes from 7 minutes to 30 seconds, that time is credited back to every affected customer on every outage, every year. This is how an in-house dispatch tool becomes a regulatory-grade reliability instrument.
Resource Availability — Process Comparison
What dispatch looked like before — and what it looks like now.
The legacy process forced operators to cross-reference five or more fragmented sources before they could dispatch a single ticket. The new process, built around how the team actually works, collapses that into a few clicks inside Nighthawk.
Resource Availability — What It Does
One application, built around how Grid Operations actually dispatches people.
One application that brings together on-call + leave calendars, real-time CADS status, mobile click-to-call, and a full audit trail — everything an operator needs to dispatch the right person, fast, without cross-checking five other systems.
Project 3 — Operator Switching Dashboard
Making switching quality measurable, peer-anchored, and safety-reinforcing.
Writing a switch order is the most consequential thing a DCC operator does. A clean order keeps crews safe; a flawed one can mean an arc flash, an unplanned outage, or a crew member in the wrong place. This dashboard measures that activity at the resolution the risk deserves.
- Workload imbalance. Most switching was being written by the same few operators — they felt overwhelmed, while other operators weren't improving.
- Metrics that drove bad behavior. Ranking operators by raw counts rewarded writing more (with high error rates), “reviewing” more (without finding errors), and revising orders that didn't need to be revised.
What AI Did on These Projects
Not a sidecar tool — a collaborator from the whiteboard to production.
Three platforms. Six months. One small team. And OG&E owns every line of code.
How We Use AI — End to End
A seven-step cycle from objective to Nighthawk production.
Align with
Stakeholders"] --> B["2
Prompt-Engineer
with AI"] B --> C["3
Analyze Data
with AI"] C --> D["4
Build Prototype
with AI"] D --> E["5
Stakeholder
Review"] E --> F{"6
Approved?"} F -- "No — iterate" --> B F -- "Yes" --> G["7
Deploy to
Nighthawk"] classDef step fill:#0070c7,stroke:#003366,stroke-width:2.5px,color:#ffffff classDef decide fill:#ffc740,stroke:#b38000,stroke-width:2.5px,color:#1a1a1a classDef ship fill:#2bb24a,stroke:#186b2a,stroke-width:2.5px,color:#ffffff class A,B,C,D,E step class F decide class G ship
Every cycle produces two outputs: a better application, and a sharper sense of how to use AI well on the next one. The process itself improves with each project.
The Journey Since October 2025
What an AI-first development pace actually looks like inside a utility.
Three substantive operational applications — plus a supporting portfolio, all on Nighthawk — delivered in roughly six months. AI is what made that cadence possible on a team our size.
Questions?
Let's Talk About What's Next
What Would You Like to Know?
Happy to dig into any of it — how we built these, what we learned, or how the ideas might translate to your DCC.