Cleaning Out the Garage
Sprint 0. Before building something new, create the space for what comes next.
Late March 2026, we had a working infrastructure, a set of established tools, and fifteen years of analytical methodology that lived almost entirely in the heads of a handful of senior people. The question we sat down to answer was not whether any of it worked well. It was whether building within it would allow us to build what we actually wanted.
The answer was no. And the work of Sprint 0 became defining what needed to be moved before anyone wrote a line of code.
***
Sprint 0 is an unusual phase in any build because, in the conventional sense, nothing happens: no features ship, no client sees anything, no revenue moves. But the decisions made here determine what everything that follows can become.
We started by assembling a small internal team to define what came next. Bill Skelly would lead the company overall, while the build was distributed across clear domains: Molly Rutledge on methodology, Kyle Braendel on execution, Joel Cochran on infrastructure, Tim Duer on commercial, and Erin Rogers managing the project. Everyone had a defined lane, a clear operating cadence kept the work moving, and escalation was expected to be fast when something blocked progress. The goal was not control, but clarity.
***
The first thing we cleared was the infrastructure.
Cleaning out a garage is something most people avoid for good reason. For a while, you can work around what's there and make do. But eventually, if you want something different, you have to open the doors, take everything out, and decide deliberately what goes back in. It is not comfortable or fun, but it is necessary if you want something different.
We faced the same choice with our existing setup: the tools, environments, and systems that worked well for what we had been, but not necessarily for what we were trying to become. So we opened the door and evaluated what needed to stay, what needed to change, and what needed to be left behind.
We worked through those decisions one by one. We migrated off Microsoft Teams to Google Meet, moved collaboration into Google Workspace, and stood up a new Google Cloud environment separate from the legacy Azure infrastructure that had supported our services business for years. BigQuery would become the data foundation.
These were not housekeeping changes. They marked the point where the AI effort stopped being something inside an existing system and became something structurally distinct from it.
The second clearing was the operating model: what standards would govern the new system and how it would behave.
So we defined them from scratch. Four of them established how the system should be built: adapting to how work actually happens rather than forcing work into predefined structures, keeping human judgment in the loop on every output, making each new capability easier to build than the last, and ensuring the sourcing behind every output remained visible.
The fifth principle was pivotal: it committed us to making the Causeway Way—our long-standing analytical methodology—an explicit quality standard embedded into every output. By doing this, we bridged the most critical divide: the space between our professional standards and what the system would automatically enforce.
This ensured our core identity remained intact even as we transitioned to a new foundation.
***
As Sprint 0 concluded, no code had been deployed. Yet, we had completed the work the garage required: emptying the space, deciding what deserved to come back, and putting everything in its proper place.
Legacy systems, inherited assumptions, and gaps in enforcement had all been examined, then either carried forward or left behind.
The team was in place.
The principles were defined.
The space was ready.
What we did not yet have was the architecture that would hold it all together.
But the garage was clear.
NEXT IN THE SERIES: The Decision That Shaped Everything Downstream. The architectural choice that determined what the platform could eventually become.