👋🏼 Hey there! Welcome to another roundup edition of Offload, a newsletter for professionals to learn how to build products and automate work with AI and no-code tools.

In this issue, you'll find:

• A complete guide to automate business proposals from call transcripts with AI

• Learn what a system prompt is (vs. user prompt) and why it matters

• Two of the most impactful things I've read this week, curated for you

• Learn to make a Simpsons version of yourself and put it into a scene

-Offload This-

🤖 Automate your business proposals with AI

People waste so much time preparing documents and presentations.

Especially those that exist only because a call happened, like proposals, briefs, or summaries…

For exemple: great sales call, solid rapport, clear needs… and then comes the dreaded task → writing the freaking proposal 😩

This teardown fixes that.

You'll learn how to turn a sales call transcript into a complete proposal. Automatically.

Here is a summary diagram of what it looks like:

But this isn’t just about sales. This same workflow works for tons of other cases:

• Briefing call → Scope of Work

• Pitch meeting → Investment Memo

• Discovery call → Client Plan

• Sync call → Internal Summary

If your meeting leads to a document, this pattern applies.

Learning to set this up takes maybe a couple of hours, but it will free you up hundreds of hours every month!

Here is the link to the post. 

-Learn AI-

System prompt vs user prompt: why AI says what it says

In the last issue, we covered tokens, the small chunks of text AI uses to process language.

Now let’s answer the next question: How does AI decide which tokens to pick next?

The answer comes down to two types of prompts:

• System prompt: it is the invisible background instructions, telling the AI what role it plays (like “helpful assistant”), how to behave, and what rules to follow.
  

• User prompt: That’s you. Your question or request in the chat.

The system prompt sets the tone and guardrails. The user prompt gives the specific task.

Together, they shape the probabilities of what the AI says next.

The AI looks at the conversation so far and asks, “What’s the most likely next token?” based on both prompts.

Think of it like sampling a survey.

Without a system prompt, it’s like asking the entire population a question.

Every possible tone, style, and perspective is in play. You might get a scientist, a poet, a conspiracy theorist, or someone’s weird uncle.

With a system prompt, it’s like setting strict quotas: "only sample from helpful customer service agents" or "only poll pirate role-players."

Mini system prompt in chats

Here’s an interesting twist: when you start with “You are a cybersecurity expert...” in your user prompt, you’re writing a mini system prompt on the fly.

It shifts the AI’s personality and focus temporarily.

So, the AI looks at the conversation so far and asks, "Given this context, what’s the most statistically likely next token?" That’s it.

The system prompt adjusts what’s "likely." You can try it in any chat, for example:

The takeaway? Both prompts shape how AI responds. Whether you’re chatting, coding, or roleplaying with a pirate bot, it’s all about how you steer the context.

-Tools I'm Using- 

1. Make.com*: a no-code automation platform that connects different apps and services using visual workflows. Useful when you want to automate complex tasks across tools like Airtable, Gmail, OpenAI, and more.
   Using score: ⭐️⭐️⭐️
   

2. Tally.so: a no-code form builder that lets you create forms quickly without design or coding skills. Useful when you need to collect data, feedback, or payments with simple, shareable forms.
   Using score: ⭐️
   

3. Airtable*: a spreadsheet-database hybrid that feels familiar like Excel but works like a lightweight backend. Great for organizing data, managing workflows, or even powering full apps when paired with automation tools. I use as a database for most of my projects.
   Using score: ⭐️⭐️⭐️⭐️

_{* Indicates links that are commissioned. You help me if you use them.}

-Reading list- 

Today I'm testing out this new section, “reading list”. It is basically a curated list of the two or three things I've read that somehow had a strong impact on me…. hopefully for you too.

1. “Breaking Down the Infinite Workday”

I recently explored Microsoft’s “Breaking Down the Infinite Workday” report. Its an eye‑opening look at how the digital era has erased the traditional 9 to 5 for knowledge workers.

Based on trillions of anonymized signals from Microsoft 365 and insights from a huge survey, the study uncovers a deeply fragmented workday.

Check out some of the scary stats:

It is impossible to be productive in these settings, and clearly an opportunity for AI solutions.

2. Sam Altman's Startup Advices

This list of 95 pieces of advice for startups from Sam Altman is 12 years old, and honestly, every single one still holds up.

I went through them all a couple of times, and it’s clear that the fundamentals of building something meaningful haven’t changed. If you are a builder, read it.

-Fun with AI-

Create a Simpson character and add it to a scene you like

This can be made for any cartoon, like Family Guy, Tin Tin, Charlie Brown, South Park … it makes sense when there is a very distinctive drawing style.

Follow these steps:

1. Copy and paste this prompt below, and attach a picture of yours

For example, here is mine (I didn’t have a better picture, sorry):

Result below. Notice how I'm not “Simpson asian”, but keep going..

2. Ask to add the character to a scene you like. Attach the scene as support. In my case, I added myself to the couch scene of the Simpsons intro and specified how.

The result:

Nerd test: there is something wrong with one of the Simpsons characters in the scene created. Do you know what it is without looking at the original pic?

_{The answer is: Homer's hair is wrong.}

-Any feedback-

Before you go, I’d love to know what you thought of today's newsletter so I can improve it for you. 🙃