Notes on reading the principles in 2026
The manifesto was adopted in 2024. Since then AI has changed how software gets built and what it costs to build it. The text stays as it was adopted, because that is what people signed. These notes read each principle with AI in the picture.
AI is the biggest trend most CTOs will see in their careers. The principle still holds. Every AI project gets the same test as anything else. Does it help the business grow and win?
Many will. Support gets cheaper, products gain features customers pay for, and teams ship sooner. Plenty won't. A chatbot added because a board member asked for one is chasing a trend. The CTO's job is to know which is which, and to say so out loud.
AI adds a new line to the budget, and it behaves differently from the old ones. Model and token costs grow with usage, seats multiply across the company, and the bill can swing month to month.
Financial discipline here means knowing the cost per task, setting limits before usage takes off, and checking that the spend returns more than it costs. A CTO should be able to tell the CFO what AI costs and what it buys.
Building the room now includes deciding what agents do. Some work moves to AI tools, and the people on the team spend more of their time reviewing, directing and deciding.
Someone on the team still owns every outcome, including the ones an agent produced. The CTO sets clear rules for who reviews AI work, who ships it, and who answers for it when it breaks.
The tools change every few months. A plan built around one model or one vendor will be out of date before it ships.
Adaptability means building so a model can be swapped out, running small experiments before big commitments, and changing course when something better arrives. The outcome the team is after stays the same while the tools underneath it change.
AI makes some corners very easy to cut. Code nobody reviewed, customer data pasted into a tool nobody approved, model output shipped as fact.
The CTO decides where company data may go, which tools the team may use, and how AI output gets checked before it reaches a customer. When an AI system makes a mistake, the company owns it, and so does the CTO.
AI can produce a lot of code very quickly. Code that nobody on the team understands is a burden the day it ships.
Sustainable here means the team can still explain, test and change what it built, whoever or whatever wrote the first draft. Speed today counts for little if the next team can't maintain the system.
Many CTOs are building again because of these tools. Prototypes that took a week now take an afternoon. That is worth protecting.
Curiosity is also a guard against burnout. A team that is afraid AI will replace them won't do its best work. The CTO can give people time to learn the tools, share what works, and enjoy it.
CTO Levels sets a level with 2 questions. How many people report up to the CTO, and how big is the budget? AI is pulling those 2 numbers apart. A small team can now carry a large budget, much of it compute.
How that should change the levels is a question we haven't answered yet.