Today we speak with Michael Dougherty — “Agile Mike” — about how a 30-year veteran of solution development and product leadership made AI a working part of his Agile coaching practice. Michael walks us through the journey from early ChatGPT curiosity, to documentation as the first real win, to the personal-passion project that built his comfort with the tools, and finally to where every Scrum Master should start: the retrospective.
From Curiosity to Toolbox: A Slow Burn That Started With User Stories
“I tried, and I thought it sucked at that. I’d rather write this by hand.”
Michael’s AI journey started in late 2022 with ChatGPT — like most coaches, he tried to make it write user stories and acceptance criteria. It didn’t work. The output wasn’t usable, and he went back to writing them by hand. What looks like a failure was actually the first lesson: not every coaching task is a good fit for AI, and the way to find out is to try. He kept experimenting in the background through 2023 and 2024 — Claude, Grok, Perplexity — while keeping personal use and business use cleanly separated. The real shift came at the Department of Homeland Security, where he sat alongside ex-Googlers, ex-OpenAI people, and an Anthropic alumnus on the team building DHS’s internal chatbot. Watching what AI could do inside a heavily-protected enterprise environment changed his frame: this wasn’t a productivity hack anymore, it was infrastructure.
The Boring, Tedious, Now Quicker: Documentation as the First Real Win
“AI has taken all those boring, tedious tasks and made them quicker, more valuable, and more fun to boot.”
The first AI practice that stuck was documentation — the part of Agile coaching nobody enjoys but every client demands. Michael had spent years writing SDLC docs from blank pages, and his co-author on Shift: From Product to People used to look at empty docs and ask for help getting started. AI removed that blank-page problem. He’d give the model the team context, the client’s documentation requirements, and a template — and get a draft to edit. He stayed in the loop as the human, bringing context, judgment, and final shape, but he no longer started from zero. The same shift happened with presentations: instead of fighting PowerPoint spacing, he tells the AI to drop content into a template and gets back a formatted deck. Small tasks, but multiplied across a coaching week, they bought back real time.
The Personal Project Trick: Find Something You Love First
“Find something you have a passion for, something you really enjoy. That made me feel comfortable with AI so I could do more.”
Michael’s strongest advice for coaches who feel intimidated by AI isn’t to start at work — it’s to build something for yourself, in a domain you care about. His own example: the Nordic Metal Tour Tracker, a personal AI agent for tracking heavy metal bands across Northern Europe (and Spain and Portugal). It plays music. It maps tours by country. It exists purely because he loves heavy metal. The trick isn’t the tool — it’s that working on something he actually cared about removed the fear and built the muscle memory. Once the comfort was there, transferring it to work became a small step instead of a leap. He references the rundown AI as a daily input that keeps him aware of how others — even people in their 80s — are using AI for everything from gardening to balanced school lunches.
A Week With Preppy Paul: Agents, Connectors, and the End of Inbox Drudgery
“Preppy Paul gives me a list of all my meetings today and the top three items I should have on each one — based on my notes, my email, my calendar.”
Michael’s current weekly routine looks nothing like his Agile coaching week from three years ago. He works at Zion Cloud Solutions as a Tier 1 Google AI partner, splitting time across three roles: about 20–30% traditional Agile project management (coaching a Scrum Master on a state of Illinois project), and the rest as an AI Growth Engineer and AI delivery lead on short 4–6 week projects with 3–5 person teams. The backbone is Google Gemini Enterprise — a platform that lets him run Claude, ChatGPT, or any major model behind enterprise protection (Model Armor, the SAIF framework). Inside it, connectors pull in Gmail, Outlook, Calendar, Confluence, Jira, Slack, Teams, monday.com, and thousands of other tools — each agent inherits only the permissions the user already has. A few practical patterns he runs every week:
- Preppy Paul — an agent that prepares him for the day’s meetings, pulling top topics per meeting from his notes, email, and calendar.
- Meeting-to-email — AI records and summarises every meeting, and he asks the agent to draft updates to specific team members directly from the summary.
- LinkedIn drafts in his voice — hashtags, formatting, and tone all match his style; he edits lightly and posts.
The pattern across all of these: the agent isn’t replacing the work. It’s removing the friction between the meeting and the next conversation.
The Retrospective: The Best Place for Every Scrum Master to Start
“Don’t have it be the same darn Jira spreadsheet of the same darn three questions every time. Go to AI and eat your heart out with it.”
When pushed for one thing a Scrum Master could try tomorrow morning, Michael was clear: the retrospective. It’s the one Scrum event where creativity is usually welcome, where format experimentation has a low cost, and where teams are already expecting something new. Ask AI for retrospective formats that fit the team’s current situation — review five or ten options, pick one, run it. The signal you watch for within a week is double: how many improvements the team generated, and how many of them actually got done. The simpler signal is the team’s reaction. If at the end someone says “that was fun” — and then you tell them you used AI to design the format — you’ve shown the team what AI is good for without a single slide or framework. It’s the smallest possible experiment with the largest possible upside.
Experiment Like a Scientist, Not a Spectator
“Don’t be afraid of it. Try whatever you can with it. Have the viewpoint of a scientist.”
The trap Michael sees most coaches falling into is curiosity without commitment — reading about AI, watching demos, talking about it, but never building anything. His framing is the opposite: be a scientist. Run experiments. Some will fail (his user-story attempt did). Some will quietly become essential (documentation did). The way to find out which is which is to keep trying, in low-stakes contexts, with both personal projects and small work experiments. The coaches who’ll be useful to their teams in 18 months aren’t the ones who can describe AI accurately — they’re the ones who’ve built enough small things to know what AI is bad at, what it’s surprisingly good at, and where the human in the loop has to stay.
About Michael Dougherty
“Agile Mike” has over 30 years of experience with solution development and product leadership, working in nearly every IT role that exists and literally hundreds of companies during his career. Michael has taught multiple Agile courses to over 1000 people, spoken at multiple events and podcasts, written dozens of blogs, and has been recently serving the US Government. He is the other co-author of Shift: From Product to People.
You can link with Michael Dougherty on LinkedIn and find more about his work at shiftingpeople.com and AP8XGlobal.com.
31st August 2026 at 10:05 CET*
BONUS How AI Took the Boring Out of Agile Coaching With Michael Dougherty*
