Insights
Our latest thinking.
Notes on AI-native software delivery, engineering transformation, and what we’re learning inside real engineering orgs.
New whitepaper · · Free PDF
The Bottleneck Isn’t the Model.
A flow-first playbook for AI adoption in software delivery: how value stream mapping, the Theory of Constraints, and the lean toolkit turn faster developers into faster delivery.
Read the summary & get the PDFShowing 7 of 20 posts
Ten agent prompts to find your delivery bottleneck this week
Where is software delivery actually slow? The evidence is already in your repos, tickets, CI, and Slack. Give a coding agent read-only access, paste these ten prompts, and compare against the benchmarks instead of guessing.
Operating model·7 minThe Constraint Ladder: where your bottleneck goes after you fix it
Fix a bottleneck and it doesn't disappear. It moves, and in AI-adopting engineering organizations it moves in a predictable sequence. Knowing the next rung before you get there is the difference between an AI roadmap and a series of surprises.
McCaul Baggett
Operating model·7 minYour work spends 95% of its life waiting. AI doesn't touch that.
Flow efficiency in knowledge work runs 5–15%: for every hour of real work, six to nineteen hours of waiting. AI speeds up the working hour and does nothing for the waiting hours, and the waiting hours are most of the calendar.
Carter Musick
Perspective·6 minYour developers are faster. Your releases aren't.
Every metric your developers can see says AI is working. Several metrics your board can see say it isn't. Both sets of numbers are telling the truth, and the gap between them is where AI budgets go to die.
Chris Sims
Modernization·6 minYour tech debt is a loan. Do you know the interest rate?
“The codebase is a mess” has never been funded. A dollar figure has. How to compute the annual interest you're paying on technical debt, and the one-page dashboard that keeps the paydown funded once you get it.
McCaul Baggett
Modernization·6 minStop ranking tech debt by annoyance. Score it like a portfolio.
Most tech debt backlogs are ranked by whoever complained loudest. Gartner's PAID model replaces that with three scores a CFO will accept, and it changes the funding conversation.
McCaul Baggett
Perspective·7 minAI displacement is real. Stop hiding behind the slogan.
“AI won't replace you. Someone using AI will.” The line sounds pragmatic. It's a dodge. If your AI strategy doesn't include the people affected by it, it isn't a strategy. It's cost-cutting with better branding.
Chris Sims
Questions, answered
About these notes.
What we write about, who writes it, and the question underneath most of it.
- Do AI coding tools actually make engineering teams faster?
- Not by themselves — the research and our field experience agree on that. Tools dropped into an unchanged delivery process produce felt speed, rework, and flat delivery metrics; the lift shows up when the system around them changes too. That gap between AI motion and AI results is the recurring subject of these posts.
- What topics does Sigao Insights cover?
- The working problems of AI-era software delivery: technical debt and what it really costs, legacy modernization, spec-driven development versus vibe coding, managing coding agents at scale, and proving AI value to a board. Plus the essays behind our eight core commitments, for the culture side of the same story.
- Who writes these posts?
- Sigao practitioners — the people running our delivery, modernization, and transformation engagements. These are notes from the field, written from what we're learning inside real engineering orgs, not a content calendar.