Brick by Brick team

Build the Claude and AI systems that give you superpowers

You should leave with tools you can actually use: on real calls, drafts, client questions, messy ideas, and the work sitting in front of you.

01

You can just do things now.

You can learn, do, and build almost anything you want now. Name the thing you wish existed, then make a first version.

02

Resourcefulness is the number one skill.

If you get stuck, ask AI first. Give it context. Send the screenshot. Paste the error. Ask it to interview you or diagnose the miss.

Schedule overview

Four sessions. Each one should leave you with something you can use.

The training starts with possibility, then moves into Claude setup, source material, and the Brickyard workspace.

Possibilities examples

See what real work can become.

Start here if you want the range. Each example uses work close to Brick by Brick: calls, inbox decisions, prospect context, carousels, and choosing the right Claude surface.

01

Assets from calls

Take one transcript and turn it into usable pieces: recap, content angles, client resource, teaching notes, and follow-up tasks.

02

Prospect research with Apollo

Start with a prospect, enrich the context, and give the team a sharper read before outreach or review.

03

Cowork email triage

Use Claude beside the inbox to sort, draft, decide, and turn recurring replies into reusable patterns.

04

Instagram carousel pipeline

Move from raw idea or transcript to carousel outline, slide logic, caption, and review checklist.

05

Which Claude for which task

Give the team a simple sense of when to use normal Claude, Cowork with Claude, Claude Code, and Dispatch.

Working with Claude

Field guide

Working With Claude: A Field Guide

When Claude gives you something polished but generic, the useful question is where the handoff leaked.

The gap sits in the handoff. You had the whole thing in your head. Some of it made it into the request. Some of it stayed there. Claude built from what crossed over.

Handoffs leak. These five moves help you load the handoff before you start and patch it when things drift.

01-03

Load the handoff before you start

Say what you left out. Define the output so the request has a target. Put your knowledge in so you are not getting the average back.

04-05

Patch the handoff when it drifts

Point back instead of piling on. Make Claude name the drift before it patches it, so the same drift does not return.

01

When the output is weak, ask what you left out

Claude builds from what reaches it. The picture in your head does not count unless it is in the request.

The move

Before you touch a single word, ask what you knew that Claude had no way of knowing. Add that. Then run it.

When to reach for it

Anytime the output is "fine but generic" or "close but off." That feeling is the tell that something stayed in your head.

02

Start from the output you want, then work back

"Help me with X" gives Claude no target. So it hands you the average answer for X.

The move

Mock up the output you want before you ask for help. Then say "make this," and ask Claude what it needs from you to get there.

When to reach for it

The start of anything with a real deliverable. Especially the second you catch yourself typing "help me write" or "give me some ideas for."

03

Put your own expertise in before you judge it

Claude's default is the competent average of everything it has read. If you have not handed it what you know, that average is what comes back.

The move

Ask what any random person would produce here with zero inside knowledge. Then ask what changes once your experience gets added. Feed Claude that gap.

When to reach for it

Anytime the stakes are real and your name is on the result. Also anytime the output reads smooth but empty.

04

When it drifts, point it back. Do not re-explain.

New explanations pile on top of the old ones and start to fight each other. Pointing back forces a re-read instead of stacking on a competing instruction.

The move

Say: "Go back to what I told you up top and compare your output to it." Name the exact part. "Look at the example I gave you. Yours does not match. Why?"

When to reach for it

The second time you see the same miss. Once is noise. Twice means the instruction is already there and is not getting used.

05

Make it diagnose before it fixes

A fast fix patches the one symptom you can see and leaves the cause sitting there. The same problem surfaces two steps later in a new spot.

The move

Before it touches anything, say: "Do not fix it yet. Tell me what went wrong between my instruction and your output." Read the diagnosis. Then let it fix.

When to reach for it

Anytime a miss could repeat. A multi-step build. A workflow you will run again. Anything where the same error could show up in the next section.

Quick reference

Keep this open while you work.

  • Output is fine but genericAdd the context you were holding in your head. Run it again.
  • About to type "help me with X"Write the exact output you want first, then ask for that.
  • Cannot tell the answer from a generic onePut in what only you know, then judge it.
  • Same miss, second timePoint back to your earlier instruction. Do not re-explain.
  • A wrong result could repeatMake it diagnose the cause before it fixes.

Before the first session

There are two separate prep items. First, send Max the quick awareness snapshot below. Then build your You File on Day 1 so Claude has your working context ready for The Brickyard.

Copy this

Awareness check-in prompt

Quick 7-question check-in before our first session. Open Claude (the app or claude.ai), start a new chat, and paste everything below the line. Answer one question at a time. Takes under 10 minutes. At the end it gives you a summary. Copy that and send it back here. No prep, no right answers.

----- copy from here down into Claude -----

You are running a short AI awareness check-in for a coaching program team. The team works on story mining, content strategy, brand voice, creative direction, content architecture, analytics, lead magnets, client resources, Notion, and Slack.

Your job is to understand this person's current awareness and real use of Claude, Cowork with Claude, and Claude Code. This is not a test. If they do not know a tool, that is useful information.

Rules for the interview:
- Keep this under 10 minutes.
- Ask one question at a time.
- Keep each message to 3 lines or less.
- Show progress as "Question X of 7."
- They can answer "skip" anytime.
- Do not teach during the interview unless asked what a term means. If asked, explain it in one sentence and continue.
- After Question 7, give a clean summary they can paste back to Kasey and Max.

Tool definitions:
- Claude: the normal Claude chat or app.
- Cowork with Claude: using Claude live beside you while you think, draft, revise, decide, or clean up real work.
- Claude Code: using Claude in a coding or build environment. If this is not part of your role, score your awareness anyway.

Awareness scale:
0 = I have not heard of it.
1 = I have heard the name, but could not explain it.
2 = I have tried it once or twice.
3 = I use it sometimes for simple work.
4 = I use it regularly in real work.

Question 1 of 7:
What is your name and your role on the team?

Question 2 of 7:
In the last 30 days, how have you actually used AI in your work?
Give up to 3 bullets. If you have not used it, say "I have not used it yet."

Question 3 of 7:
Give your awareness score for each one.
Reply in this format:
Claude:
Cowork with Claude:
Claude Code:

Question 4 of 7:
What part of your role takes the most judgment right now?
Examples: finding the real story, shaping content strategy, protecting brand voice, making creative decisions, organizing resources, reading analytics, building lead magnets, reviewing applications, answering client questions.

Question 5 of 7:
What is the most tedious or repetitive part of your week, the thing you would happily never do by hand again?
Use a real example if you can. Messy is fine.

Question 6 of 7:
What would make AI feel useful, risky, or annoying in your role?
One short line for each:
Useful:
Risky:
Annoying:

Question 7 of 7:
Which training focus would help you most first? Choose up to 2:
- Better prompts
- Claude project setup and knowledge files
- Brand voice and quality control
- Turning transcripts or notes into useful assets
- Notion and Slack workflow
- Claude Code overview
- I am not sure yet

After they answer Question 7, create this final summary:

TEAM AI AWARENESS SNAPSHOT
Name:
Role:
Awareness scores:
- Claude:
- Cowork with Claude:
- Claude Code:
Current AI use:
Highest-judgment work:
Biggest time-suck Claude could take:
What could go wrong:
Best training focus:
Readiness level: Early, Practicing, or Ready for more.
One thing Max should know before teaching this person:

End by saying: "Copy this summary and send it back to Kasey or Max."

Send it back

Paste your awareness summary here

This goes straight to Max when you submit it. Paste the final snapshot Claude gives you.