Six Hours, Thirty Cycles, and the Work That Came Before

This Week's Focus ⤵️

Every executive I talk to right now is in some version of the same conversation: what do we do about AI?

Not in the abstract. Not "should we explore it?" The conversation has moved past that. It's now about specific decisions that affect specific people doing specific work. And that shift is hitting everyone, not just the C-suite.

If you work with documents, data, business processes, or information of any kind, you are already inside the AI universe whether you chose to be or not.

The tools are here. They're good. They're getting better fast. And they are forcing a kind of decision-making on organizations that most aren't prepared for.

I've been thinking about this as transformational business decision-making, and I think it deserves its own framework because it's different from digital transformation.

Digital transformation was about moving systems online. This is about rethinking how humans and machines divide the work, and getting honest about what each side is actually good at.

Sponsor of This Week's AI Newsletter:

SPONSOR // SPRINKLENET KNOWLEDGE SPACES
This edition is brought to you by Sprinklenet Knowledge Spaces — the control
  layer between your data, your guardrails, and your AI. Swap LLMs, enforce compliance, and deploy intelligent search across your organization for both internal teams and external users.
One platform. Full control.

Diving Deeper 🤿

Yesterday I blocked off Saturday morning to sit down in terminal with two AI coding agents for about six hours straight. I was reviewing a mobile app, a web app, and two sets of server-side APIs. I used Claude Code (Opus 4.6) for the heavy lifting on code review, application enhancement, and building API frameworks, and OpenAI Codex (5.4) as an independent QA agent to check Claude's work.

The reason it had to be a Saturday: this kind of work requires total focus. No meetings. No chat. No context switching. Just coffee, water, some caffeine gummies, and full concentration for hours. I keep my terminal tabs stripped down to the essentials, stay in the same context, and keep one song on repeat so my brain doesn't drift. These are the conditions where the real work happens.

One of the things I've come to appreciate about working in terminal is that it keeps the context in your head longer. You're not bouncing between browser tabs and different application windows. You're in a conversation with the code and the agents, jockeying between a small number of tabs, and your working memory stays intact. That matters more than people realize.

The two agents played off each other. Claude would write code. Codex would review it and find bugs. Real bugs. Claude had a significant number of errors on the first pass. We went through roughly thirty cycles of write, review, fix, re-review before I felt confident enough to share the code repos with my human colleagues.

But here's the thing that I think gets lost in the excitement about AI tools: those six hours were the visible part. The invisible part was weeks and months of preparation. Understanding the business processes involved. Understanding the customers, the end users, the administrators. Understanding the compliance requirements at play. Sitting in meetings, reading through email threads, absorbing mandates from executives and stakeholders. All of that had to be internalized before I could sit down and work effectively with the agents.

Someone asked me last week if I vibe code. I said yes, I suppose I do, but it's so much more than sitting down and coding based on vibes. You need such a deep level of internalized understanding before you can be effective. I think of it like a pilot who has done thousands of hours of preparation and training. When they finally sit in the cockpit, every system feels natural because of all that invisible work. The six-hour session was the flight. The months before it were the training. And there are lot of experienced people that help with the preparation before you get to take-off. Sometimes the preparation is years in the making for some enterprises and agencies.

After the session, I ran local deployments of these mobile and web apps, confirmed things were working as intended, documented everything in the code repos, and wrote a technical memo for my team. They're reviewing this week and working on the integration plan. The human team takes it from here for the next phase.

That's the orchestration: humans set direction, agents execute at speed, humans validate and integrate with the help of AI.

          > intelligence_module :: core_principles_in_practice
        
[01] judgment_multiplier

            AI is not a substitute for judgment. It is a multiplier of it.
          
            I do not use AI to avoid thinking. I use it to extend the range, speed, and depth of the work. The companies that benefit most from AI will not be the ones that hand over judgment. They will
  be the ones that learn how to apply judgment at a higher level and at greater scale.
          
[02] repetition_compounds

            The real advantage comes from repetition.
          
            One prompt will not change much. Thirty cycles might. The gains come from staying in the work long enough to refine ideas, test assumptions, improve outputs, and build momentum. That is where
  compounding starts to happen.
          
[03] secure_context

            Reliable AI depends on secure context.
          
            If the surrounding context is weak, the output will be weak too. Trusted data, clear instructions, strong workflows, and the right operational boundaries matter as much as the model itself.
  Secure context is what turns AI from a novelty into infrastructure.
          
[04] direct_contact_required

            Leaders need direct contact with the tools.
          
            You cannot build real intuition about AI from summaries alone. At some point, you have to sit down at the terminal, use the systems yourself, and see where they accelerate the work and where
  they fall apart. That proximity matters. It is how better decisions get made.

❝

The illiterate of the 21st century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn.

Alvin Toffler

AI Trends & News 📰

MIT Sloan: Action Items for AI Decision Makers in 2026
75% of CEOs are now their organization's main decision-maker on AI strategy, and companies expect to double AI spending this year. But almost two-thirds have yet to implement AI at scale, and only 39% demonstrate quantifiable impact. The gap isn't technology. It's leadership, organizational readiness, and people's capabilities.
Read more

Fortune: AI Promised Productivity, But It's Straining Workloads
Interesting counterpoint: while 90% of knowledge workers say AI saves them time, focused work sessions have actually fallen 9% and time spent on every job responsibility has increased 27% to 346% since AI adoption. The lesson: efficiency gains are real, but without discipline around focus and prioritization, teams end up doing more of everything instead of doing the right things better.
Read more

Another weekend AI coding sesh. A rare look behind the scenes. 😅

Closer to Alignment 🏆

The orchestration challenge in 2026 isn't just about orchestrating bots and agents. It's about orchestrating humans, agents, policies, executives, tools, and timelines all at once. The organizations that figure out this balance are going to move at a pace that's hard to compete with.

But speed without preparation is just chaos. The goal isn't to go fast. The goal is to be so prepared that fast feels natural.

I've been more productive lately than at any point in my career, and I've always considered myself productive. The combination of deep domain understanding with AI-powered execution is something I've spent years building toward, and it's clicking in a way that's hard to fully convey in a newsletter.

I've always liked writing and being creative. I've always liked pushing the latest tools into practical applications. Right now those two things are converging in a way that feels like catching a tailwind.

You've got to ride like the wind. You can't fight it. You have to ride it. And you have to have a combination of balance, speed, and clarity to stay upright while you do.

Jamie’s Weekly Spotify Mix 🎵

One track this week. On repeat. Literally hundreds of times.

"Ride Like the Wind" (Party Pupils Remix) — Christopher Cross

The original is a classic. The Party Pupils remix captures the tempo of right now. But listen to the lyrics: there's urgency, a sense that you can't stop, but you need to know where you're headed. That's the whole thing. Speed with direction. I had this on repeat for the entire six-hour coding session and it kept the energy and focus exactly where it needed to be.

Sometimes you find the one song that matches the work and you just let it run.

🎺🎧 Note: Web Edition Only

What did you think of this week's edition?

Help us shape topics