Arcos Control Tower and Arcos Field: A Platform Built Around the Storm

A bag of screwdrivers

Erin Gold, a Slalom consultant on strategy, and I turned the kickoff’s sticky-note work into a service blueprint: four phases, from planning through demobilizing, and for each one what the command center and the field roles do, hand off, and rely on. Seen that way, the gaps between the tools were the problem.

View larger (opens the full image)

The insight that reframed everything: users didn’t need a better bag of tools. They needed a system that understood the storm restoration process and gave them what they needed next, instead of making them hunt for the wrench in a bag of screwdrivers.

Every product was built as a separate tool, and storm restoration is a sequence handed from one role to the next. From the blueprint and the rest of what I learned in my first four to five months, before any convoy tracking designs, I recommended a different direction. Not a modernized UI. A workflow-native platform, one that follows the actual sequence of a storm response and surfaces the right information and actions at the right moment in that process.

View larger (opens the full image)

Making the future tangible

My job then became turning that into a direction that product managers, engineering leads, and company leadership could see and believe in. Evidence alone doesn’t create alignment.

I developed the concept of user lenses, a framework for how the platform should adapt its experience to a person’s role in the storm process. A field crew supervisor in Arcos Field needed different information surfaces than a command center coordinator in Control Tower, even when both were working the same event. Lenses became the conceptual spine of the vision for both products.

It started as Mark 1, an early vision of the Command Center. To make it concrete, I built a working prototype in Lovable. Using AI-assisted prototyping as a leadership tool was a deliberate choice: get to something tangible fast, and use it to drive real conversation instead of abstract deck review. I worked in constant dialogue with the product managers and the design team, so the prototype reflected shared understanding, not only my reading of the evidence.

AI-assisted prototyping wasn’t about speed for its own sake. It was about compressing the distance between insight and stakeholder alignment, making the vision real enough to pressure-test.

The same rhythm held through delivery. On one planning milestone, working closely with Product Leadership, the research and design work delivered UX strategy up to 9 weeks ahead of planning sessions and set aside 2 to 4 weeks of structured discovery to check a feature’s value with utility users before development began.

Three marks of the Command Center

The Command Center is where the back office watches a storm. I created every Control Tower prototype, and the Command Center went through three marks. Mark 1 grouped tools by storm phase in its navigation. Mark 2, the testable prototype, put a full-screen restoration map behind floating panels, with a storm and restoration bar along the bottom, and leaders from the Storm Response Advisory Committee tested it. Mark 3, the near-final design, was built in three versions: one by Oswaldo Vazquez, one by Ales Karoza, and one by me. I led the design delivery.

Mark 3 is built for density. Response progress leads the screen. Every resource request sits in one table with a status chip on each row. The map layers outage areas, convoys, and crews, and the storm and restoration bar keeps the numbers that matter in view whatever the panels above are showing. Financials and timecards get their own tab.

This is a protected imageClick to unlock protected images

A note on what these captures are. The final designs stayed with Arcos when I left. The screens here come from a rebuild I made afterward, from my original prototype and the last design files I had, running on fictional data.

One screen, four phases

The blueprint’s phases show up directly in the product. Mark 1’s navigation names them Plan, Mobilize, Restore, and Reconcile, and each opens an overview of its own: crews by status, a panel chosen for that phase, then the crew list. During Mobilize the crew map takes the second panel; during Restore it’s notifications; during Reconcile, a skills breakdown of every crew on the event.

	Plan	Mobilize	Restore	Reconcile
Mark 1 Command Center	View larger (opens the full image) Planning Phase Overview	View larger (opens the full image) Mobilization Phase Overview	View larger (opens the full image) Restoration Phase Overview	View larger (opens the full image) Demobilization Phase Overview

Two cuts, one convoy

Crew check-in was the problem the redesign set out to solve. Before storm restoration can begin, external crews have to be checked in at a staging site, and that process was slow, error-prone, and still largely paper-based. I prototyped a check-in workflow.

Then executive and product leadership cut scope. Twice. The first release shipped on schedule in August 2025, about ten months after I joined, on both sides at once. The desktop app, then called Storm App, had convoy tracking, destination management, and login, so command center teams could see where inbound external resources were and prepare for their arrival. Arcos Field had login, account creation, event invites, convoy creation, and convoy tracking, so a crew could join its event and create its convoy on the phone. The product took the name Control Tower after that release. Leaders from the Storm Response Advisory Committee and internal Customer Success users tested convoy tracking end to end.

Convoy tracking still had to make sense from both ends. The back office filters convoys by status and time to arrive; the crew’s phone walks one convoy from Mobilize to Inbound to Arrived.

The check-in process was still manual and off-system when I left. The first release didn’t ship the thing the redesign was meant to fix. That’s a real limitation, and I own it as part of the story.

Convoy tracking was meant as a low-disruption first step, the start of moving customers off an older legacy solution onto the new platform without overwhelming them with a full feature suite on day one. That movement was still lagging when I left. The vision hadn’t changed, though, and the user lenses concept was still the north star.

Watching storm mode happen

From August to October 2025, after that first release, I spent two and a half months on-site with American Electric Power, visiting 14 sites across six of their operating companies and spending time with both command center staff and field workers, roughly 8 to 10 people per site. The fieldwork was planned for Mobile Workbench, one of Arcos’s legacy products, and I folded in questions about time entry that became the Timecards work. I led the research, with Erin Gold and Steve Herring, both from Slalom, alongside me. None of it happened in a conference room. We sat with command center operators, went to staging sites, and watched what the workflow looks like under pressure.

Command center operators were running 4 monitors with 8 different applications open at once. Field workers kept a paper notebook, their “bible”, because it was more reliable than any digital tool when pressure was high and connectivity was spotty. The fieldwork didn’t start the reframe, but it sharpened it: what we saw went into the Control Tower and Arcos Field personas and the Control Tower service blueprints.

What shipped, and when

In production, August 2025: convoy tracking, destination management, and login on the desktop, and login, account creation, event invites, convoy creation, and convoy tracking in Arcos Field, all designed by me and launched together. January 2026: user profile management on both phone and desktop, which Apple’s App Store required. After I left, Arcos Field Timecards, Control Tower Timecards Management, and Control Tower Rosters reached production in September 2026, and the Command Center was released to beta.

Holding the longer arc

The most important thing I did on Control Tower and Arcos Field wasn’t the research, the prototype, or the design work, though all of those mattered. It was keeping a clear and credible picture of where the platform needed to go, even when organizational constraints meant it couldn’t get there all at once.

The AI-assisted prototyping piloted during this project also outlasted it. By August 2025, product managers and designers were using Lovable on their own to turn product needs into tangible examples for leadership review, faster than before. That capability didn’t exist at Arcos when I arrived. Building it was part of the job.