Practical writing on AI in software development, team practices, and business results.
All articles ↓More clarity. Stronger teams.
Episode five of 3 AImigo on entering IT, the differences between building a project, learning, getting hired and getting paid, and how to show what you understand when working with AI agents that carry out assigned project tasks.
Cases from my practice, three lines next to every number: measured on what, adjusted how, where it stops. Lead time 2.5x shorter, a product in ~2 weeks.
History in this browser
All articles are listed below. Search and topic filters require JavaScript.
Found: 21 articles
Episode five of 3 AImigo on entering IT, the differences between building a project, learning, getting hired and getting paid, and how to show what you understand when working with AI agents that carry out assigned project tasks.
People and teams
Episode four of 3 AImigo on hiring engineers who work with AI: work trials, assessing decisions, responsibility and choosing the next role.
People and teams
Episode three of 3 AImigo on the cost of a completed task, QA rework, model selection and the systems agents need to do useful work.
AI adoption
The second 3 AImigo episode on early checks, human attention, internal tools, and the people who change how a company delivers.
AI adoption
The first 3 AImigo episode explores why teams use AI differently and how they can share methods that have been shown to work.
People and teams
We already have enough capability and long-horizon runtime. What I want now is faster models and a better working relationship between operator and agent.
Working with agents
Loop engineering automates execution, not authorship: a loop pursues a supplied goal but cannot choose that goal or decide when it should change.
Working with agents
Operational radar for CTOs: what to adopt now, what to pilot, what to assess, and what to keep out of production in an AI-ready SDLC.
AI adoption
Three eval loops for AI features, AI tooling, and agentic/MCP workflows: capture layer, golden set, grader in CI, and vendor-drift detection.
Working with agents
AI made writing code dramatically cheaper, but moved the cost into review, CI/CD, and governance. While this tax stays invisible, it keeps growing.
Impact and measurement
Pilots usually fail not technically, but because after the demo there is no owner, production template, metric, and go / no-go decision.
AI adoption
AI assistants need rules next to code: what is allowed, how to reason, and which checks run before merge or deploy.
Working with agents
Notion Agents turn documentation from a folder of pages into a managed layer: policy answers, processes, gap audits, and onboarding without constant senior routing.
AI adoption
Spec-Driven Development connects business, product, and engineering through a specification, Definition of Ready, and automated implementation conformance checks.
Working with agents
The Annex III and Annex I dates are corrected against the EU timeline; the remaining legal advice in this article has not been reviewed.
AI adoptionCorrected the EU timeline: Annex III applies from December 2, 2027, Annex I from August 2, 2028; remaining legal advice was not reviewed.
A CTO self-assessment: operating model, governance, hiring, and metrics. If fewer than 6 of 12 are yes, AI is already in use, but the system is not assembled yet.
Impact and measurement Cases from my practice, three lines next to every number: measured on what, adjusted how, where it stops. Lead time 2.5x shorter, a product in ~2 weeks.
Impact and measurement
96% of companies buy AI and get PoCs without production impact. 4% build an operating model: owner, CI/CD, review, metrics, and explicit noes.
Impact and measurement
AI is already changing hiring: test thinking, work with an agent, and ownership, not boilerplate-writing speed.
People and teams
A format for CTOs of 30-150 engineer companies: interviews, review/CI audit, baseline, governance gaps, roadmap, and a kill / rescue / rollout decision.
AI adoption
A CTO checklist: how to distinguish a production contractor from a seller of PoCs and presentations before signing.
AI adoptionTry another word or clear the filters.
if this sounds like your team →
If these articles describe your engineering organization, a paid diagnostic - one to two weeks as the standard scope - maps the bottlenecks and returns three options for the next step, each priced.
See work formats →or send a short email with one link and one question:
[email protected]