Change Leadership - Advisory Practice
Standing up the technology tooling is the easy part. Changing how your engineers decide, and keeping that judgment theirs, is the engagement.
RAND studied why AI projects fail and found the causes upstream of the technology. Five root causes, drawn from interviews with the engineers and data scientists who lived them. The one named most often was organisational: people misunderstanding, or failing to communicate, what problem the AI was meant to solve. By some estimates over 80% of AI projects fail, twice the rate of IT projects that use no AI.
That is the view from outside. Here is what it looks like on an engineering train.
What actually stops it
It reads as more work. Most organizations carry the memory of an initiative that promised leverage and delivered documentation. A new practice competes with that memory before anyone has made the case.
The people who could lead it have the least capacity. The architects with enough context to shape how AI gets used are the same ones carrying the critical path. Asking them to take this on asks them to drop something else.
Explaining costs more than doing. An engineer who could walk six people through a system ships a lower-resolution version instead. The walkthrough costs two days. The shipping costs one. Expertise stays put for rational reasons.
Credit goes to whoever delivers first. Any programme asking people to share more argues against an incentive nobody has changed.
How we work
We supply the form and support structures. Your people supply the content.
An architecture group we worked with recently could not agree on what their real problem was. Forty-five minutes in a room together produced nothing. So we sent them a one-page form for defining a problem well, and stayed out of it. They worked it themselves that evening. Four days later the architect who had rejected the whole framing read their statement to the train and defended its scope. Not one line of it was ours. That is what made it hold.
The test is what survives
Abandonment kills more of these than rejection ever does. Something gets built. Enthusiasm carries it a month. Two quarters later it is stale enough to be worse than nothing.
So anything we help you stand up has to hold itself up. That is a design constraint from the first week, not a hope for the last.
The engagement
Find where AI is already being used, and by whom
Establish what good looks like, in your language, agreed by the people who have to meet it
Coach the practice into work already in flight rather than alongside it
Build maintenance into the design so the thing survives the enthusiasm
Leave the definition, the standard and the facility with your peopled
We provide the process, support strucutures, and tehcnical guidance; what we leave behind is a team that can do this without us.
Tooling is the easy part. The work is changing how engineers decide, and keeping the judgment theirs.
Change Leadership - Advisory Practice