bsTechTeam — Blum & Schaeffer
Student
Journey
A progressive curriculum spanning two days — from first contact with AI to commanding a team of agents. Every student is the President. The agents work for you.
Day
1
2
Foundation
Early stage
What is AI? — Day 1 Block 1
What is an LLM?
A Large Language Model is software trained on massive amounts of text. It doesn't think — it predicts. Every word it generates is a statistically informed guess based on everything it was trained on. Claude is an LLM.
What is a context window?
The context window is the LLM's working memory. Everything in the current conversation — your messages, Claude's responses, any files or instructions — lives inside it. When the window fills up, older content falls out. Nothing persists on its own.
What is a token?
Tokens are the units LLMs actually read and write. Not words — chunks. "Hamburger" might be two tokens. A space is often its own token. Token count determines how much fits in a context window and how much an API call costs.
First Conversations — Day 1 Block 2
Iterative chat
Each message is a complete, self-contained request. You write a prompt, Claude responds, you refine and send again. The goal is learning to iterate — treating each reply as a draft to improve, not a final answer to accept.
Prompting basics
Good prompts are specific, give context, and state the desired output format. Vague prompts get vague answers. Think of prompting like briefing a smart contractor — the clearer your brief, the better the work.
The % command
The % command is how you address a specific agent. "% [agent name] do X" routes the task directly. It's the core of the bsTechTeam command protocol — you are always the President giving direction, never the one doing the work.
Gate 1 Student can hold a productive conversation with Claude and use the % command.
Gate 1 | How to save multiple conversations?
You can hold a focused, productive conversation with Claude and direct it with intent.
You understand what's happening under the hood — tokens, context, prediction.
Now that you are comfortable with the basics, how do conversations persist across sessions?
You understand what's happening under the hood — tokens, context, prediction.
Now that you are comfortable with the basics, how do conversations persist across sessions?
Workflow Mastery
Middle stage
Claude Architecture — Day 1 Block 3
Projects & threads
A Project is a persistent workspace. Threads are conversations inside it. Projects remember across sessions. Threads are where the work happens. Keep them organized — context bleed is a real problem.
Skills + RAG
Skills are reusable instruction sets that give Claude a specific capability. RAG feeds Claude your own documents instead of relying on training alone. Skills tell Claude how to behave. RAG gives it what to know.
Project Instructions
Standing orders for your project. Claude reads them first, every time. Define your voice, rules, and mission here. In the military framework, Project Instructions are the project's Base General. They don't take orders.
Command & Agents — Day 2 Block 1
Military framework
You are the President. Your agents are officers. They don't think — they execute. The rank structure maps directly onto how Claude works. This is the mental model that makes everything else click.
Build agent roster
An agent is Claude configured for a specific mission. Your roster is the list of agents active in your project — their name, purpose, rank, and use count. You build it. You own it.
Project Files
The documents and memory your project carries across sessions. Drop your SOPs, rosters, and references here. Claude works from them automatically. In the military framework, these are your Colonels. Agents are your Captains.
Gate 2 Student understand agents and now has a live agent roster.
Gate 2 | Manual maintenance of agents is tedious
One agent isn't enough. You can feel it. The roster works but the tracking is manual, the work is useful but too much copy-pasting, and the operation is only as big as you can hold in your head.
You're ready for infrastructure.
You're ready for infrastructure.
Infrastructure
Graduate stage
AI Created Landing Page — Day 2 Block 2
Build a Landing Page
Create a simple landing page with the help of AI coding agents and attract potential users.
Connect to Web and Berserker Mail
Now the landing page is built we connect your web presence and Berserker Mail for seamless integration.
Deployment & Infrastructure — Day 2 Block 3
Domain Setup and App Deployment
You've been running agents inside Claude. MindStudio lets you build those same agents as standalone apps. Zapier connects them to your business tools. Your roster just got a deployment pipeline.
MindStudio + Zapier
You've been running agents inside Claude. MindStudio lets you build those same agents as standalone apps. Zapier connects them to your business tools. Your roster just got a deployment pipeline.
Gate 3 Utilization and ROI Student runs a AI assisted outbound sales environment with 12 week support.
Gate 3 Utilization and ROI
The manual work is behind you. Everything you built by hand — every logged use, every project file — is now running automatically. You built it. You own it. You command it. Time to deploy field ops.
What students leave with
After Day 1
AI-literate Operator
- Understands how LLMs work
- Holds productive AI conversations
- Uses the % command fluently
- Thinks of themselves as the President
AI-literate Operator
Understands how LLMs work
You'll know what's actually happening when Claude responds. Prediction, not thinking. Tokens, not words. Context windows, not memory. That foundation changes how you prompt, how you plan, and how you lead.
Holds productive AI conversations
You don't just chat with Claude — you direct it. You know how to frame a task, iterate on a response, and recognize when the context window is working against you. One turn at a time, purposefully.
Uses the % command fluently
% is your command syntax. It's how the President issues orders. You don't type requests anymore — you issue directives. That shift in posture is small on the keyboard and enormous in outcome.
Thinks of themselves as the President
You don't climb ranks. Your agents do. You set the mission, deploy the officers, and hold the command layer. That's not a metaphor — it's the operational reality of how this system works.
By Lunch Day 2
Agent Commander
- Manages projects and threads
- Builds and maintains an agent roster
- AI Power User
Agent Commander
Manages projects and threads
You've moved past single conversations. You run projects — each with standing orders, persistent files, and organized threads. Your Claude environment is structured, scoped, and mission-ready.
Builds and maintains an agent roster
You know how to configure an agent, name it, assign it a mission, and track it in your project file. Your roster is a living document. It grows as your operation grows.
AI Power User
You leverage Claude's full capabilities to enhance your productivity and decision-making. You understand how to use advanced features and integrations to streamline your workflow.
After Day 2
Infrastructure operator
- Deploys and manages Custom Landing Page
- Applies LLM work to increasing outbound sales
Infrastructure Operator
Deploys and manages Custom Landing Page
Your agents have identities, missions, and a permanent home outside Claude's native interface. You deployed them, your workforce is real and it runs on your infrastructure.
Applies LLM work to increasing outbound sales
You use Claude's language model to analyze customer data, identify opportunities, and create targeted sales strategies. Your approach is data-driven and results-oriented.
Doctrine
% means: agent do
agents execute. command thinks.
good agents get promoted
you are always the President