The noise-overwhelm problem
The volume of AI advice available is a problem in itself. Too many prompt frameworks and acronyms to count. Workflows built around developer tools designed for software teams rather than people running a business. Conflicting recommendations, constantly updated information, all from people whose context potentially has almost nothing in common with yours.
Alongside all of it: a growing sense that learning to work with AI properly is just another item on an already crowded list of things to do. More to research and figure out before any benefit arrives.
The sprint is a guided cut-through. Five days. Thirty minutes each. Everything in it is tested, sequenced, and built for the way a small business or solo operation actually runs. No acronym frameworks. No developer tooling. No prerequisites beyond a Claude Pro account and a note-taking system you already use.
Five days. Each one compounds.
Each day is a 30-minute 1:1 session. Guided, not recorded. The session names what you need to understand, you build something real during it, and you leave with one thing to complete before the next day. Each day feeds directly into the next.
Day 1
Why the output is inconsistent, and how to change that.
The model is completing a pattern from whatever you give it. Vague input returns averaged output: not a flaw, but the mechanism. The first session closes that gap: from correcting output to directing it.
Day 2
Ending the re-explanation problem.
Every AI conversation starts from zero. The model knows nothing about you unless you tell it, so you re-brief at the start of every session: friction that compounds across everything you do. The defaults start aligning with how you work rather than requiring correction every time.
Day 3
Getting the AI out of the output.
The hedges, the summarising preambles, the generic structure: the model writes toward its statistical defaults unless constrained otherwise. Style is not inferred; it has to be specified. You identify the patterns that conflict with your voice and build the instruction set that eliminates them.
Day 4
From re-uploading to building.
Every session can only hold what you put into it that day: when it ends, so does the context. A project fixes that, holding your brief, decisions, and background in a persistent container that every relevant session opens with already loaded. From that point, the project carries everything forward.
Day 5
A system, not a stack of sessions.
By day five, personalisation is running, voice is specified, and active work has persistent context. The difference between isolated setups and a system that accumulates is infrastructure: a knowledge vault the AI reads from and writes to. By the end of it, you have something that compounds rather than resets.
What changes
The first day or two will feel like learning rather than doing. That is the draw: a short period of investment before the system starts to behave like it knows you. By day three the personalisation layer is working and the AI's default behaviour has noticeably shifted. By day five you have built something that continues to compound the longer you use it.
The shift from “I use AI” to “I work with AI” is not gradual. It has a moment. This sprint is designed to create that moment in about two and a half hours across a week.
Who this is for
Small business owners and solopreneurs who are already using Claude but not compounding the value of it. People who have read some things, tried some things, and still feel like they are working harder than the tool warrants. Or people who are not yet sure where to start and want to begin correctly, not experimentally.
If you are looking for a certification, a general AI survey, or a course that requires restructuring how you work before it delivers anything, this is not it. The sprint assumes thirty focused minutes a day for five days and a willingness to build something real in that time.
If the five days are useful
On day five we review what you have built and where you want to take it. If a deeper build makes sense, I will show you what we could accomplish in 90 days.
The next step is the Claude Cowork System: weekly sessions over 90 days where you design and embed an AI operating system calibrated to your specific work, tools, and context. What the sprint introduces, the System builds out fully.
There is no obligation either way. The five days stand on their own.
Common questions
Do I need prior Claude experience?
Some familiarity helps. If you have used Claude a handful of times, that is enough. The sprint starts from foundations. If you have not used Claude at all, Anthropic offers free foundational training at anthropic.skilljar.com, worth completing before day one, but not necessary.
How much time does it take each day?
Thirty minutes of focused work per day. Each session has a clear start and end point.
Is this self-paced or facilitated?
Guided 1:1. Each day is a live 30-minute session with Peter. The intimacy is the point: the guidance is calibrated to where you are starting from, your specific business, and what you are trying to get off your plate.
What do I need?
A Claude Pro account and a note-taking system you already use. If unsure, give Obsidian a go.
What happens after the five days?
On day five we review what you have built. If a deeper engagement makes sense for your context, we discuss it then. If not, you leave with a working foundation you can continue developing independently.