Selected Work
From Content Editor to AI-Enabled Operations Builder
How I used AI-assisted development to turn a new operational role into a connected system for reporting, visibility, knowledge, and enablement
I resisted AI at first. When Kiro showed up, I didn't know what it was or what "vibe coding" meant, and I figured it was a tool for developers. My background is content, communication, and digital learning, so it didn't seem like it had much to do with me. A year later I was running a six-report communications cadence, maintaining a searchable operating guide for an established Content Operations team, and shipping interactive learning games for onboarding. This is the story of how that happened, what I built along the way, and what it taught me about using AI to make an organization work better.
The Business Challenge. A newly established sprint-based operating model needed a supporting ecosystem for reporting, stakeholder visibility, process measurement, documentation, and training.
My Approach. I applied AI-assisted development, structured requirements, and operational communications experience to design and build a connected set of tools and workflows.
The Outcome. Six recurring reports, preparation time cut from more than an hour to under ten minutes, self-service stakeholder visibility, a searchable operating guide of 100+ pages, and four reusable interactive learning experiences.
Where I Started
For most of my time at Delta I was a content editor. I spent years managing customer-facing content across Delta.com, the Fly Delta app, and airport digital platforms, which meant promotions, shopping experiences, and a lot of careful work inside AEM. Before Delta I spent decades building and running websites, intranets, and internal communications programs, first at Kennesaw State University, then at Aprio, and then at ABM Industries. So I wasn't new to technology or to communication. I was just new to the idea that I could build my own software.
Then, almost by accident, I started building games. I'd been curious about Kiro, and one afternoon I sat down and described a small game to it, mostly for fun. A concept that would have sat in a notebook for months turned into something I could play before dinner. Then I built another one. What got me wasn't that AI could spit out working code. It was how fast an idea became something real that I could poke at, break, fix, and bend into something that was actually mine. I'd felt that same rush on the early Web, when anyone curious enough could just start making things. I hadn't expected to feel it again.
The Role I Inherited
Meanwhile, my job was changing under my feet. In January 2026 I moved from content authoring and editing into the Communications Lead role for Delta's Content Operations team, an established group supporting dozens of stakeholders across the organization. The team was moving to a structured sprint-based workflow, and my new responsibilities covered sprint management, stakeholder communications, reporting, intake, and operational documentation. A new operating model like that creates an opportunity to build a more complete supporting framework around it, and that framework was mine to shape.
Five areas stood out as the place to start:
- Reporting designed around the sprint model, drawing on data the team's workflow system already held
- Visibility that lets stakeholders see where their requests stand without having to ask
- Measurement that shows how the new process is performing, beyond counting completed requests
- A centralized, maintainable home for operating knowledge
- Enablement that gives people practice, not just something to read
I could have built all of that by hand, and in a way that's what a traditional communications lead would have done. It also would have eaten my whole week, every week, with nothing left over for improving anything. I needed a different engine.
Learning by Building
So I made Kiro part of how I learned the new job. I didn't take a course on AI, and I wasn't handed a playbook. I learned the way I learn everything, which is by making something, hitting the limit of what I knew, and figuring out how to get around it. The games were my practice field. By the time I turned the same approach toward real operational problems, I'd already learned the habits that mattered most.
The biggest one was to stop treating AI like a vending machine. Early on I'd describe something loosely and hope for the best, and the results were about as good as that sounds. Over time I leaned on structure instead: written requirements, design documents, implementation plans, task lists, and iterative testing. If I could define the behavior I wanted, break the work into pieces I could manage, and check the results, the output got a lot better. Who's going to read this? What should be left out? What does a good version look like? Those questions were mine to answer, and they were the same ones I'd been asking as a communicator for thirty years.
To be clear about who did what: I owned the problem definition, the requirements, the information architecture, the workflow design, the testing, and how each tool fit into the way the team actually works. Kiro did a great deal of the typing, and it let me build at a speed I couldn't have managed alone. But every system on this page reflects what I know about this organization and the people who use it.
The lesson I keep coming back to is simple. Build right away, then make it yours. Describe an idea, get a working version, open it up, tear into it, and go again. The distance between "I have an idea" and "I have something I can use, test, and improve" has never been shorter.
What I Built
What started with arcade games grew into a set of connected systems, each one built to answer a problem the new role handed me. Each has its own case study if you want the details:
- A six-report communications cadence spanning weekly, biweekly, monthly, and quarterly reporting for different audiences, generated through an AI-assisted workflow with defined rules for what to include, what to leave out, and how it should read. Read the case study →
- Stakeholder visibility platforms that turned recurring status questions into self-service answers. Read the case study →
- Operational dashboards that make complicated AEM configurations readable at a glance, with production and staging side by side. Read the case study →
- A quarterly process maturity report that gives leadership and the team a structured, AI-assisted look at how the sprint model is performing. Read the case study →
- The Content Operations Process Guide, a searchable operating guide of 100+ pages across eight categories that consolidates distributed process knowledge into a centralized, maintainable resource. Read the case study →
- Four reusable learning games for onboarding and training, built around different kinds of knowledge: sequencing, recall, decision-making, and process comprehension. Read the case study →
Around those sit a whole collection of smaller utilities, trackers, and one-off apps built around the way I work. Some of them only I will ever use. That's fine. Not everything needs a launch.
Impact at a Glance
The Result
The clearest number is time. Recurring reports that took more than an hour to prepare now take less than ten minutes, and that's with six of them on the calendar. But the time wasn't really the point. Because preparing a report stopped being a chore, I could put the hours I got back into the parts that need a person: deciding what a report should say, talking to subject-matter experts, and noticing what the data was trying to tell us.
Some of the gains are harder to put a figure on, and I'd rather say so than invent one. Stakeholders can see where their work stands without asking anyone, which cut down on manual status requests, but I don't have a before-and-after count. The Process Guide gives new teammates and stakeholders one place to look. The process maturity report gave leadership something they never had, a way to watch the new operating model grow up. And the reports started traveling past their intended audience. Leaders several levels above me noticed them, sent feedback, and asked what else we could show them. That told me the work wasn't administrative overhead. It was becoming a real source of organizational visibility.
What I Learned About AI and Organizations
The biggest lesson wasn't really about AI. Every system on this page started with a plain question about people. What does this reader need to know? Where would a new teammate look for this? What decision is this report supposed to help someone make? The AI made it practical to act on the answers quickly. It never supplied the answers. When I skipped that thinking, the output came back generic. When I did the thinking first, the tool could build almost anything I described.
I also learned that human judgment has to live inside the system, not around it. One early design decision was to keep reporting focused on operational outcomes rather than individual attribution. I wrote that principle into the reporting requirements so every report after it served the business purpose it was meant for. Nobody told the AI to care about that. I did. Oversight is a design choice you make at the start, and it's much cheaper there than at the end.
Finally, speed changes what's worth trying. For a learning team, that's especially exciting. Interactive practice, scenario-based learning, knowledge checks, job aids, and prototypes used to take serious development effort. Now I can explore them in days instead of months, put a working version in front of the people who would use it, and find out quickly whether the idea deserves more of my time. The cost of being wrong dropped, so I'm wrong more often, earlier, and in smaller ways. I'll take that trade every time.
I'm still learning every day. But I've learned enough to be convinced that the people who get the most out of AI won't be the ones who let it do the work for them. They'll be the ones who use it to power up their own creativity, judgment, and know-how.
What This Demonstrates
- AI Enablement
- Moving from skeptic to practitioner by building real tools, and bringing a structured, requirements-first method to AI-assisted work.
- Operational Excellence
- Recognizing that a new operating model needs matching systems for communication, measurement, documentation, and training.
- Workflow Automation
- Turning recurring manual compilation and analysis into repeatable, governed workflows.
- Knowledge Management
- Capturing subject-matter expertise in a searchable guide that is built to be maintained, not just published.
- Learning Design
- Translating procedural and decision-based knowledge into reusable interactive practice.
- Self-Directed Growth
- Learning a new discipline in the middle of a new role, by building things people actually use.