Spec-first AI workflows
I’m spending time on the parts of AI-enabled delivery that need to get sharper as implementation gets cheaper: specifications, acceptance criteria, review loops, and clear ownership boundaries.
Now
What I am focused on right now
This is the most time-bound page on the site. It is where I keep track of the work, questions, and experiments that feel current rather than settled.
Current focus
What I’m spending time on
Written handoffs and decision records
A current focus is how teams reduce delivery noise with stronger defaults around planning, written handoffs, and decision clarity rather than more process theatre.
Architecture proposals that earn their cost
I’m paying more attention to proposals that add coordination or maintenance overhead, and whether that cost is actually justified by the delivery reality around them.
Shorter notes, published more often
I’m trying to write shorter, more concrete notes that make delivery observations easier to carry forward while the questions are still live.
Current experiments
Where AI actually belongs in the workflow
The experiment I keep coming back to is not whether AI can write implementation quickly. It can. The more useful question is where teams need tighter structure once that speed exists: specifications, review quality, ownership boundaries, and the written record around the work.
So this is less about tool novelty and more about operating model design. I’m interested in what changes when AI becomes part of delivery architecture rather than a shortcut layered on top of existing habits.
Open conversations
Open conversations
These are the conversations I am most likely to say yes to at the moment.
Teams redefining engineering standards around AI
I’m interested in conversations about how engineering standards, team design, and leadership expectations change once AI starts affecting throughput without changing accountability.
Delivery systems that reduce coordination noise
I’m always keen to compare notes on planning, architecture, and developer experience changes that quietly improve throughput without adding more process burden.
Writing, talks, and selective collaboration
I’m open to thoughtful writing, podcast, speaking, and advisory conversations where the overlap is software delivery, technical judgement, leadership, and AI in practice.