In partnership with
Every post you read here starts the same way: with my voice.
I talk through the idea out loud, in Portuguese, the way I would explain it to a colleague. A typical recording turns into three to four pages of text. Across my voice drafts so far, the average is about 1,600 words.
What happens between that recording and your inbox is today's tip. It's my real workflow, step by step: what I do, what AI does, and why the final decision is always mine.
A five-minute read with 30 years behind it
A post may take you five minutes to read. What's behind it took 30 years.
The stories are real. They come from my own career, from what I see my colleagues need, and from what the companies I work with are asking for right now. When a customer asks me "Is this OK?", or a colleague sends me just "Hi", that's where a post begins.
AI wasn't in those rooms. It doesn't have those memories, and it doesn't know which story matters this week. So two things are always mine: the ideas, and the final decision.
But I also believe everyone today should be using AI to help with the rest. I do, every week.
My workflow, from voice note to published post
Here is what happens, in order:
I record a voice note. In Portuguese, about 1,600 words. No structure yet, just the story and the lesson.
The draft is saved as is. It goes into my private GitHub repository, word for word. That way I can always compare what I said with what was published.
AI writes the English version in my voice. I wrote down how I want to sound: correct, clear English, short sentences, a Brazilian voice, no marketing tone. The AI follows that guide, and checks technical facts against the official documentation.
AI prepares everything around the text. The newsletter draft with title, subtitle and preview text, the images, and a review document with a checklist for me.
I review. Is this what I mean? Is the story true and complete? Would I say this sentence out loud? I ask for changes until the answer is yes.
I decide. I approve the text, upload the images, pick the ad, and schedule the post.
AI publishes the rest. Tracking links, the posts for LinkedIn, Instagram and YouTube in English and Portuguese, and, starting with this post, a Portuguese email for readers who prefer it.
I listen. Replies, feedback and results go back into the next idea.
The AI notetaker that gets the hard words right
Your AI is only as good as what you feed it. Feed it a transcript with your product names, acronyms, and numbers spelled wrong, and you spend the afternoon hand-fixing the output.
Wispr Flow Notetaker uses your dictionary and calendar, so product names, acronyms, numbers, and uncommon names come out spelled right. It pops up when your meeting starts and captures Zoom, Google Meet, Teams, and Slack huddles with one click, no bot joining the call.
Then your meetings show up inside Claude or ChatGPT through the built-in connector. No copy paste. Try Notetaker free on Mac and Windows.
Everything lives in one place, my repository. Here is what it looks like:
Even the workflow itself is written down there, so the AI follows the same steps every time:
And every draft is measured against the final post. My recordings ranged from about 800 to more than 3,000 words:
Where AI really helps
Details we forget, or never knew. An exact setting name, a date in the documentation, a limit that changed last month. AI finds them fast. I still check what matters.
Publishing. Many tools now accept automation through connectors. The most common standard today is MCP, the Model Context Protocol. It lets an AI assistant work directly inside a tool: my newsletter platform, my Google Drive, my code repository. The heavy part of the work, the copying, formatting and filling in fields, is done for me.
Small details that make a big difference. The subject line. The preview text. The right image size for each channel. Alt text for people who use screen readers. A short link that is easy to type. Many of these looked unimportant to me at first. They aren't. A single phrase or the wrong image format can change how many people read a post.
Images, for people who aren't designers. I'm not one. The diagrams in my posts are generated from code, in my brand colors, and I approve every one.
You don't learn all of this at once. You learn it post by post, and your knowledge grows with every one.
Ask for feedback, and use it
The other thing I do is ask for feedback, from colleagues and from anyone who reads what I publish. And when it comes, I take it seriously.
Recently, two colleagues and friends gave me tips. One was about how I use AI. The other was about the language itself.
If English isn't your first language, you know this. Even if your English is very good, even fluent, some things stay with you for life: an accent, a way of building sentences. You use fewer idioms, or you don't even know some of them exist. Native speakers notice, even when they can't say exactly why.
I looked at their feedback carefully, found it very useful, and made it part of my routine. Some of it I had already started, but not at the level they described. Their tips took me further than I was going on my own.
That's also why every post ends with a short note on how I write. I want you to know exactly how it's done.
This Friday's tip: three steps
1. Write down your own workflow. Pick something you do every week: a report, a status email, a presentation. List each step, and mark it: me or a tool.
2. Keep the "me" steps, and hand one "tool" step to AI. The ideas, the stories, the decisions and the final review stay with you. Choose one heavy, repetitive step and let AI do it this week. Then review the result.
3. Ask two people for honest feedback, and change one thing. Not to feel good. To get better.
Your Friday homework
Reply to this email and tell me: which step of your work would you hand to AI first? And which one would you never give away? I read every reply.
If you know someone who still does all the heavy work by hand, forward this to them.
Prefere receber os e-mails em português? Escolha aqui.
Prefer to listen? Two AI hosts debate this post, one for it and one skeptic, on our podcast, We The People, in Tech or Not: The Debate: listen to the episode.
How I write: every post starts from my own notes and experience. I use AI tools to help draft and edit, and I review every word before it goes out.
Disclaimer: The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of Microsoft. The author is a Microsoft employee.
