Skip to content
  • There are no suggestions because the search field is empty.

Recommended LLM Training for Our Team

Staff tips for learning to work with LLM models in productive internal ways.

 This will be a live article with frequent updates throughout the year.
 The key is skills. Wheither formal or informal if you want solid big picture results you will need to boild down some details first. Often the llm can do that but if you rely too much on it you get slop.
 
Entry Level
 
 
 
I was shocked to have this tool make very complex Azure Cloud updates I would have hated to figure out for me.  An extremely advanced (also well documented) interface.
 
Beginner
 We will add more entry level videos soon that cover more get up and running now with this simple idea (see the 2026 new year growth coach idea below)


Overview of Business Changes      Dec

The True Test      Dec

Pro Prompting      Dec

Best Prompting Video      Dec

Choosing the Right Model Video      Dec

Next Level Detail on Model Choice      Dec

Tip:  Prompting is no longer just writing a few sentences.  The more complex the issue the more you want to do bite sized chunks, prompt those, build that into a project and in the end most advanced projects will have a skills document that outlines exactly what attributes are most important for this project.  A resources doc for important links, images and things like that.  A general outline of the process and where that lands within a goals outline.

So skills + resources + process + goals = A great modern project.  Can be done in a single simple prompt but more or less most prompts should lean in this direction and complex ones need this at the bare minium.

Pretty boring but about as simple as you'll ever find regarding agents      Dec

Part 2      Dec

https://www.anthropic.com/learn      Dec

https://www.anthropic.com/learn/claude-for-you      Dec

https://anthropic.skilljar.com/      Dec

https://academy.openai.com/      Dec

Obsidian Introduction

Obsidian is a free phenomenal simple way to organize your AI growth process, prompts, data in a format that can be instantly dropped into any LLM.

Real World Use Cases:  We vibe coded all of these without using the slightest technical knowledge other than posting the code to the web.  All 6 months ago with a much less effective Claude model.  Our customers and staff use these right now.

https://movezen360.com/movezen-owner-funds-income-expense-calculator/

https://movezen360.com/rental-vacancy-cost-calculator/

https://movezen360.com/investment-rental-property-financial-analysis-projection/

This was built in December from the ground up as a prototype to several challenges we face now (I built these programs years ago on ancient code) using the most modern framework in less than 2 hours human time. https://mz-proto.vercel.app/
(though setting up Claude code took a couple hours and was tedious, it did the entire process for me)

Moderate


Skills:  The thing to keep in mind with this is even if you see a formal method of creating and using skills, it is not all that formal and you can easily replicate similar functions with basic text that's the power of LLM's.

Building Skills      Dec

Advanced Prompting      Dec

Claude Code and MCP Overview      Dec

 
 
 
Github has always been crucial to any software process but it's more important than ever now because with vibe coding you can publish a lot more software, especially as a team and that requires organization.
 
 
Advanced
 
Up to now I cautioned that LLM's were a waste of time for most people.  Now I think it's a major mistake not to aggressively implement them in your workflow. 

I would say the same (too raw waste of time) for Agents and MCP now (for average people).  However, that will probably no longer be the case at the end of 2026.  So things change insanely fast on this front.  At some point if you don't get on the train it will leave the station and you will never be advanced similar to web development tinkerers in the late 90s.  Learn while it's infantile because you will not understand anything soon unless those foundational ideas are locked in. This process is moving about 5x faster than the proliferation of the web.
 
 
 
 
 
Training and fine tuning : In a couple years we'll all have local models on our machines and they will be highly trained on our data and info but more importantly they will maintain "persistence" a lot more, meaning it will have a much better understanding of you as an individual and will grow with you over time where old models are used to update their replacements so this info will be useful in time.
 
Claude Code from the Creator:  While this is highly technical, the concepts would work very well on a basic LLM on your local machine and if you just "do your best" to match what they are doing.  That's the power of LLM's is that it should do a lot of the guiding for you, or as they often allude to in these videos figure it out for you.
 
In this realm effort matters more than expertise.
 
 
 
 

 

Bonus 2026 Idea.  Hire a Coach

 


Give me a simple but through step by step method to use gpt as a 2026 goal and progress project where I submit info up front, schedule monthly or quarterly follow ups in which you would quiz me on KPIs, and provide with progress reports. Also small nudges through the year as I work on new things.

This is great but this is for my coworkers who are going to want something a little easier to digest.

Got it. You want “simple system my coworkers won’t hate,” not “founder-brain OKR manifesto.”
Here’s a cleaned-up, coworker-friendly version you can basically drop into an email, Notion, or onboarding doc.
––––––––––––––––––
HOW TO USE CHATGPT AS YOUR 2026 GOALS COACH
(Simple, copy-paste friendly version)
––––––––––––––––––
Purpose:
Use ChatGPT to help you set 2026 goals, track progress, and get honest feedback every month or quarter.
You only need three things:
  1. Clear goals
  2. A short “2026 goals brief” you reuse
  3. A simple monthly / quarterly check-in prompt

 
  1. Pick a few 2026 goals

Choose 3–5 goals for the year. Keep them specific and measurable.
For each goal, write:
  • Goal name
  • Why it matters
  • What “success” looks like by 12/31/2026
  • 1–3 KPIs (numbers) you’ll track
Examples: revenue, deals closed, projects shipped, hours saved, health metrics, etc.
This does not need to be fancy. Simple bullets are enough.

 
  1. Create your “2026 goals brief”

This is a short text block you’ll paste into ChatGPT each time.
Template:
“2026 GOALS BRIEF – [Your Name]
Context:
  • Role: [your role / team]
  • Main focus in 2026: [1–3 bullets]
Goals & KPIs:
Goal 1: [Name]
  • Why it matters: [1–2 sentences]
  • Success by end of 2026: [clear outcome]
  • KPIs:
    • [KPI 1]: target = [number]
    • [KPI 2]: target = [number]
Goal 2: [Name]
  • Why it matters: …
  • Success by end of 2026: …
  • KPIs: …
[Add more goals if needed]
How I want ChatGPT to work with me:
  • Ask for my numbers (KPIs) first.
  • Don’t accept vague answers like “it’s going fine.”
  • Give me clear feedback: where I’m on track / off track.
  • Suggest 3–5 concrete priorities for the next month or quarter.”
Save this somewhere easy to find (notes app, OneNote, etc.).

 
  1. Use this prompt for MONTHLY check-ins

Once a month (for example, on the 1st), open ChatGPT and paste:
“MONTHLY 2026 CHECK-IN – [Month]
Use the 2026 goals brief below.
  1. First, ask me for any KPI numbers you need, goal by goal.
  2. Then ask me 3–5 reflection questions about the month.
  3. After I answer, please:
  • Summarize my progress vs each goal (on track / at risk / off track).
  • List my key wins and key problems.
  • Suggest 3–5 clear priorities for next month.
  • Point out where I might be fooling myself or avoiding something.
Here is my 2026 goals brief:
[PASTE YOUR 2026 GOALS BRIEF]”
Then you just answer the questions and numbers it asks for.

 
  1. Use this prompt for QUARTERLY reviews

Once per quarter (early Jan / Apr / Jul / Oct), paste:
“QUARTERLY 2026 REVIEW – Q[1/2/3/4]
Use the 2026 goals brief below.
  1. Ask me for this quarter’s KPI numbers and any big events/wins/losses.
  2. Compare this quarter to the previous one.
  3. Then:
  • Tell me where I’m building real momentum.
  • Tell me where I’m stuck or repeating the same problems.
  • Suggest what I should stop, start, and double down on.
  • Give me a simple 90-day plan with 3–5 main actions.
Here is my 2026 goals brief:
[PASTE YOUR 2026 GOALS BRIEF]”
Again, answer its questions and let it do the analysis.

 
  1. Use quick “nudge” prompts during the year

Any time you start thinking about a new project or feel off track, you can paste:
NUDGE FOR A NEW PROJECT
“NUDGE CHECK – NEW INITIATIVE
Context: [what I’m thinking of doing]
Related 2026 goal(s): [which ones]
Rough time/energy cost: [estimate]
Please:
  1. Tell me if this supports or undermines my 2026 goals.
  2. Suggest 1–2 simple ways to measure success.
  3. Suggest the smallest test version I could run before committing fully.”
Or if you just feel behind:
“NUDGE – PROGRESS REALITY CHECK
Using my 2026 goals brief, ask me 5 short questions to test whether I’m actually moving toward my goals this month. Don’t let me stay vague.”
Paste your goals brief after that, answer the questions, and use the output to adjust.

 
  1. Remember to schedule it

To keep this simple for yourself:
  • Add 1 monthly calendar event: “ChatGPT 2026 Monthly Check-in”
  • Add 4 quarterly events: “ChatGPT 2026 Quarterly Review”
Each time, you just open ChatGPT, paste the relevant prompt + your goals brief, and respond to what it asks.





Business Version for Ideas


This is great but this is for my coworkers who are going to want something a little easier to digest.

You’re basically asking to turn me into a slightly judgmental OKR coach. Acceptable.
Here’s a simple but genuinely robust setup you can use all year.

 
  1. Decide what you actually care about in 2026
Keep it tight so this doesn’t turn into a spreadsheet from hell.
Pick 3–6 big goals, for example:
  • Business growth (revenue, doors under management, etc)
  • Systems / leverage (automation, data, ops)
  • Personal performance (health, deep work, etc)
For each goal, define:
  • Goal name
  • Why it matters
  • Success by 12/31/2026 (clear outcome)
  • KPIs (3–5 per goal, max)
  • Update cadence for each KPI (weekly / monthly / quarterly)
You can literally draft this as a short text table or outline.

 
  1. Build a “2026 GPT briefing” you’ll reuse every time
You want one master chunk of text you can paste in at the start of a session so I stay calibrated.
Example structure you can fill in later:
  • Context: who you are / what business looks like
  • 2026 Goals: list with a sentence each
  • KPI list:
    • Goal 1 – KPI A, KPI B, etc (definition + target)
    • Goal 2 – …
  • Constraints: what you are NOT allowed to suggest (e.g. “don’t suggest adding 20 meetings”)
  • Style: how you want me to behave
    • “Quiz me hard on numbers, don’t accept vague answers.”
    • “Flag anything where I’m kidding myself.”
You save that text somewhere handy (notes, Obsidian, whatever).
Whenever you start a “2026 session” you paste:
“Use the following as my 2026 goals brief for this session:” + that block.

 
  1. Define the standard MONTHLY check-in script
Your monthly message to me can be basically the same every time.
Template:
“MONTHLY 2026 CHECK-IN – [Month]
Use my 2026 goals brief below.
  1. First, ask me for all KPI numbers you need.
  2. Then ask 3–5 short reflection questions.
  3. Then:
    • Summarize progress vs targets
    • Call out risks / where I’m off track
    • Suggest 3–5 concrete priorities for next month
    • Give me 3 uncomfortable questions to think about
Here is my current 2026 goals brief:
[PASTE BRIEF]”
Then I:
  • Ask for KPI values in a structured way (Goal 1, KPI 1, etc)
  • Ask reflection questions like: “What moved the needle most this month?”
  • Produce a short “Progress Report” with:
    • On-track / off-track per goal
    • Key wins
    • Key failures / stuck points
    • Focus list for next month

 
  1. Define the QUARTERLY deep-dive script
Quarterly check-ins should be slightly harsher and more strategic.
Template:
“QUARTERLY 2026 REVIEW – Q[1/2/3/4]
Use my 2026 goals brief.
  1. Ask me for quarterly KPI numbers and any major events.
  2. Compare this quarter to the previous one.
  3. Give me:
    • Where I’m compounding well
    • Where I’m stalling or lying to myself
    • What I should kill, double-down on, or delegate
    • A simple 90-day plan with 3 main bets
Here is my 2026 goals brief:
[PASTE BRIEF] + [PASTE quarter’s raw notes if you have them]”
That’s it. Same script every quarter.

 
  1. Decide your actual cadence (don’t overcomplicate it)
Minimal but solid rhythm:
  • Monthly check-in: 1st or 2nd of each month
  • Quarterly deep-dive: early Jan / Apr / Jul / Oct
You can:
  • Put it in your calendar as “GPT 2026 Review – paste template”
  • Or ask me later to set up automations-based reminders in here (monthly & quarterly), and I’ll configure the recurring prompts that kick off those same scripts.

 
  1. Standard format for KPI answers during check-ins
When I ask for KPIs, answer in a clean structure so I can parse it tightly, for example:
Goal 1 – Business Growth
  • KPI: Doors under management
    • Target: 400 by Dec
    • Current: 275
  • KPI: Monthly MRR
    • Target: $X
    • Current: $Y
Goal 2 – Systems & Automation
  • KPI: Hours saved per week
    • Target: 25
    • Current: 8
etc.
The clearer your numbers, the more precise I can be when I smack your hand about them.

 
  1. How to use me for “small nudges” during the year
Whenever you start or consider something new, use a trigger phrase, e.g.:
“NUDGE CHECK – NEW INITIATIVE
Context: [short description]
Related 2026 goal: [which one]
What I’m thinking of doing: […]
Time/energy cost: [rough]
What I want from you:
    1. Tell me if this obviously conflicts with my 2026 goals
    2. Suggest how to measure success
    3. Suggest the smallest test version of this”
I’ll respond with:
  • “This supports goal X / undermines goal Y”
  • Suggested KPI(s) and check-in cadence
  • A smaller version you can try before you commit real time
You can also run a quick nudge like:
“NUDGE – It’s [date], I haven’t looked at my 2026 goals in a bit. Ask me 5 questions that will force me to confront whether I’m actually moving toward them.”

 
  1. Create a one-time “2026 Control Panel” message
When you’re ready, send me something like:
“Let’s set up my 2026 goals system.
Here is my 2026 goals brief:
[PASTE FULL BRIEF]
From now on, when I say:
  • ‘Run monthly 2026 check-in’ → use the monthly script.
  • ‘Run quarterly 2026 review’ → use the quarterly script.
  • ‘Nudge check’ → use the new-initiative nudge script.”
I’ll treat those phrases as shortcuts in future chats so you’re not re-explaining the universe every time.

If you want, your very next step can be: write that 2026 goals brief in rough form and paste it in. I’ll tighten it, define missing KPIs, and propose a concrete monthly / quarterly cadence based on how aggressive you actually want this year to be.


Bonus


The voice part is pretty spot on haha but I think ai agents have a long way to go so not quite my opinion.  For example, no agent is handling work orders (other than the same highly scripted models that have worked for years aka if this send this vendor.  If no response in 8 hours send to this series of vendors until accepted.)  in 2025 or even 2026.

what exactly is "my voice"?

Short answer: it’s the way you think out loud when you’re not trying to impress anyone.
Longer, but still digestible:
Your voice is practical, blunt, systems-oriented, and allergic to fluff. You like clarity, numbers, and levers you can pull. You get impatient with abstract motivation talk unless it turns into something operational fast. You don’t mind being challenged as long as it’s logical and useful.
More concretely, your voice tends to look like this:
  • Direct and plainspoken. Short sentences. Minimal buzzwords.
  • Focused on outcomes and constraints, not vibes.
  • Comfortable saying “this isn’t working” without drama.
  • Slightly dry, occasionally wry, but not jokey or performative.
  • Prefers “here’s the system, here’s how to use it” over inspiration.
When you explain things to coworkers, you usually:
  • Reduce complexity instead of showcasing it.
  • Give people permission to keep it simple.
  • Emphasize consistency over intensity.
  • Frame tools as helpers, not magic.
So if I rewrote that doc “in your voice,” it would:
  • Cut about 20–30 percent of the words.
  • Use fewer headings and more straightforward instructions.
  • Swap “reflect” language for “check, adjust, move on.”
  • Sound like something you’d actually send internally without cringing.
If you want to test it, say: “Rewrite this for coworkers in my voice” and paste any section. I’ll make it sound like it came from you on a focused day, not from a consultant who bills by the paragraph.

Give me an example on the topic of ai agents in 2025

Here’s a clean example of the same topic explained two ways. First is “generic AI explainer.” Second is your voice.
Same facts. Very different feel.

GENERIC / CONSULTANT AI VOICE
“AI agents in 2025 represent a significant shift toward autonomous systems capable of planning, executing, and optimizing workflows with minimal human intervention. These agents can integrate with existing software tools, reason over complex tasks, and continuously learn from feedback. Organizations should explore AI agents to drive efficiency, reduce operational overhead, and unlock new strategic capabilities.”
Sounds impressive. Says almost nothing.

YOUR VOICE (MoveZen internal style)
“AI agents in 2025 are basically software employees that don’t get tired and don’t need Slack messages to stay on task.
Instead of using AI just to answer questions, agents actually do things. They can watch a system, notice when something changes, and take action without waiting for a human to click a button.
Example:
An agent can monitor AppFolio for a maintenance request, check vendor availability, send the work order, and follow up if nothing happens. No one has to remember to do it.
Important reality check:
Agents are not magic. They’re only useful when the process is already clear. If the workflow is messy, the agent just automates the mess faster.
In 2025, the smart move is not ‘use AI everywhere.’
It’s:
  • Pick boring, repeatable work
  • Define the steps clearly
  • Let agents handle the middle so humans can focus on judgment calls
If a task still needs constant human interpretation, it’s not ready for an agent yet.”

That’s the difference.
Same topic.
One sounds like a conference keynote.
The other sounds like something you’d actually send to your team without apologizing.
If you want, I can:
  • Rewrite the 2026 goals system doc in this exact voice
  • Create a one-page “AI agents at MoveZen” internal explainer
  • Turn this into a short training or SOP format for non-technical staff