From the Google Sheet to the AI Toolkit

In 2014, I managed a 24-person co-located agile team at an agency in Gainesville, Florida. We were building MakeMyDeal, a real-time negotiation platform for AutoTrader's car dealerships. The team was the agency's largest ever. We had dedicated QA, a Scrum Master, designers, front-end engineers, back-end engineers, and a product owner from the client side.

My tools: Google Sheets, Pivotal Tracker, a physical Kanban board, and a lot of standups.

In 2026, I built a skill library of 340+ reusable consulting workflows, a personal knowledge system, a portfolio website, and a suite of interactive tools. The team: me and an AI agent.

The output is comparable. The team size is not. The interesting part is figuring out what that means.

What transferred directly

The fundamentals of product thinking did not change. Breaking work into small, testable slices. Prioritizing ruthlessly. Shipping something real before debating something theoretical. Knowing when to cut scope and when to hold the line.

The muscle I built managing that 24-person team, the daily rhythm of standups, retros, demos, backlog grooming, all of it trained me to think in systems. How does work flow through a team? Where do handoffs create delay? What information does each person need to do their job without waiting?

Those questions are identical whether the team is 24 humans or one human and an AI agent. The vocabulary changed. The thinking did not.

Story writing is a perfect example. In 2014, I spent hours in rooms with engineers and designers, breaking features into vertical slices, writing acceptance criteria, negotiating scope. The format was the same one I teach today: who is the user, what are they trying to do, how do we know it works.

With AI, I still write stories the same way. The difference is speed. What took a half-day workshop now takes thirty minutes of iteration with an agent that can hold the full context of the system in memory.

What did not transfer

The biggest gap: social capital.

Leading a 24-person team meant reading the room constantly. Sensing when someone was disengaged. Knowing which engineer needed detailed specs and which needed a sketch and freedom. Navigating the politics between the client's product owner and our agency's account team. Managing up, down, and sideways simultaneously.

None of that applies when you work with AI. The agent does not have feelings. It does not need motivation. It does not get frustrated when the requirements change for the third time. This makes some things easier and some things harder.

Easier: no coordination overhead. No waiting for someone to finish their task before you can start yours. No scheduling conflicts. No misaligned incentives between departments.

Harder: no one to challenge your assumptions. No designer pushing back on your flow because they talked to users you did not. No engineer saying "that approach will be fragile" based on experience you lack. The diversity of thought that comes from a balanced team is genuinely hard to replicate.

I compensate by building structured review into my workflow. I run my own PRDs through a skill that checks for gaps. I use different prompting strategies to simulate the designer lens, the engineering lens, the business lens. It is not the same as having those people in the room. But it catches more than I would catch alone.

The ratio shift

Here is the number that sticks with me: on the MakeMyDeal team, I spent roughly 30% of my time on product work and 70% on coordination. Standups, planning, stakeholder updates, retros, 1:1s, cross-team syncs.

Now, the ratio is inverted. I spend maybe 20% of my time on coordination (mostly with clients and collaborators) and 80% on the work itself. Discovery, strategy, content, shipping.

That inversion is the real story of AI tools for product managers. The promise is not "AI does your job." The promise is "AI removes the overhead so you can actually do your job."

What this means for teams

I am not arguing that solo-plus-AI replaces a balanced product team. It does not. For complex, high-stakes products with real users, you need humans with different expertise, different perspectives, and different instincts.

But I am arguing that the team structures we inherited from the pre-AI era need to change. A 12-person squad where 4 people are in coordination roles that AI can compress is not efficient. It is legacy.

The teams I want to build next look different: smaller, with higher individual leverage, where AI handles the mechanical work and humans focus on judgment, relationships, and creativity. The PM still writes the story. The designer still talks to users. The engineer still makes architecture decisions. But the translation layer between them, the status updates, the spec documents, the handoff artifacts, gets dramatically thinner.

We scaled up teams to handle coordination complexity. AI lets us scale them back down.

The Google Sheet is gone. The thinking behind it is not.