Better prompts, agents who make decisions, strategies that lose consistency, branding in the AI era, stop being assholes. Signals to anticipate how the world is changing in terms of how we work, sell, and decide.
Ten signals collected by reading industry sources, talking to clients in session, and testing them on my own work.
Signals to anticipate how the world is changing in terms of how we work, sell, and decide. I’ve turned them into little guides to help us get started on the week, before the week gets to us.
1. How to Write Prompts That Get the Right Results the First Time
The most reliable formula for speaking with an LLM is also the oldest: person + task + context + format . It’s the same structure recommended in Claude’s official documentation and in the OpenAI guides.
Example, tested yesterday with an operations client:
You’re an operations consultant. Turn your B2B customer onboarding into an SOP that a junior can read in 5 minutes. Format, numbered list with owner, input, output, and time for each step.
The perfect prompt doesn’t exist, but a clear brief does. When you write a prompt, you’re delegating, and if you couldn’t explain the task to a colleague, AI won’t do any better. The quality of the output isn’t in the syntax, it’s in the clarity of the thought that precedes it.
At MAKE PROGRESS®, we call it clarity before velocity . Without initial clarity, each iteration amplifies confusion instead of reducing it. I saw it yesterday in a session: a founder rewrote the same prompt 14 times in an hour. We stopped everything, defined the brief on paper in 8 minutes, and the fifteenth attempt produced the right output.
2. How to stop repeating the same old things over and over again
If every conversation with the AI starts from scratch, you’re wasting the best part of the tool. ChatGPT, Claude, and Gemini all have the blueprints ; you can store persistent instructions, reference documents, and memories that persist between sessions.
Learn to use this: a custom instruction , a system prompt that the model reads before each of your questions. Set up well, it replaces most of the context you currently repeat by hand.
I see the same mistake in companies with 8 to 800 people but zero processes. Every Monday, the same thing is explained verbally. Every Wednesday, the same mistake is corrected. Every Friday, they vow that next time it will be different, and it never is. The problem isn’t the explanation, it’s the lack of a place to document the decision.
Write once, reuse a hundred times. That’s what a custom instruction does, that’s what an SOP does. Same logic, different objects, same leverage.
3. How to Hire Your Next Fellow Agent
The technical definition of an AI agent is a system that autonomously executes multi-step workflows, deciding intermediate steps based on an end goal . This is the definition used in Anthropic’s paper Building Effective Agents .
The more you automate, the more your judgment matters about what you decide not to automate. Human work focuses on where judgment is needed: strategy, architecture, new problems the model has never seen. Everything else falls down.
Two good news hidden in here.
First: if you have clear ideas, you can delegate a number of things you wouldn’t even dream of today, and take back pieces of time you thought were wasted.
Second, you don’t have to shut down your business to get started. You need a list of the processes you’re already running out of habit. Automating is an opportunity to simplify and speed up.
Otherwise, you’ll find yourself more available and equally busy, doing activities with less value than before.
4. How to protect yourself from blue-collar wage inflation
Counterintuitive prediction but supported by two measurable trends:
progressive automation of white-collar work, with OECD estimates of around 27% of office roles exposed by 2030
If you put the two together, those who know how to do things with their hands—those who repair, install, and build—will see their wages rise. Those who have spent ten years perfecting a desk job will find themselves squeezed between automation and global competition.
For you, the owner of a small Italian company, the sign is that the plumber will cost more than your junior marketing manager within 24-36 months.
Where in your company are you already paying more for skills you’re afraid of losing?
5. How to understand that the strategy is no longer working
A strategy rarely stops working overnight. First, it becomes inconsistent in execution. The numbers come in, but not as regularly as before. You close the month, but one in three fails. You raise your forecast, but you believe in it less and less.
In an unstable system, increasing execution produces more instability. Pushing harder on a strategy that’s losing consistency is like turning up the volume on a dead radio. The MAKE PROGRESS® method starts right here; before accelerating, it restores order to the system.
Many managers describe their inconsistency as a distinguishing feature. Our clients are all different, we create tailor-made projects . This phrase sounds good and protects against confrontation, but it almost always hides a different problem: you still don’t know who you’re really working for.
If your clients come from opposite worlds, with mismatched budgets, goals, and expectations, the real question is why so many different types of people turn to me? In many cases, the answer isn’t a winning positioning, it’s a lack of alternatives within their search radius. You’re the easy choice, not the right one.
Don’t confuse inconsistency with adaptability. Agility is intentional, measured, and reversible. Emergency is reactive, dispersive, and wears down the team.
6. How to know if you’re making the right decision
When you have two options and they both seem sensible, the problem isn’t the analysis. You already have the information you need. The problem is that each option optimizes something different. One protects speed, the other control. One maximizes scale, the other quality. One protects cash flow, the other relationships.
There is no right decision, there is a choice of what you are willing to sacrifice.
A practical exercise, I do it every Monday morning with founders in coaching. For each option, write in one sentence what you’re optimizing and what you’re giving up. If you can’t complete the second part, you haven’t yet understood the decision; you’ve just looked at it.
This is one of the reasons why so many founders get stuck on the same choice for weeks. They look for the option that doesn’t require giving up. It doesn’t exist. Every strategic yes is a no to something else, often implicit, often unspoken within the team.
7. How to Brand When Everyone Writes with AI
More posts, more newsletters, more captions. Everything faster, everything more consistent, everything more forgettable. Most companies that have discovered AI are producing more content, not better content.
Your audience has changed. Today, you write for two types of readers:
Human who evaluates , as always
LLM who decides whether to advise you , when someone asks ChatGPT who can help me with problem X
This second dynamic has a name, GEO (Generative Engine Optimization), documented in the Princeton and Georgia Tech paper . It’s no longer just SEO.
Humans remember if your story is true. LLM records whether your position is clear, repeated, and consistent wherever you’re cited online. If you publish ten generic posts a month, both will forget you.
This game isn’t won by producing more, it’s won by telling a better story and repeating it in a recognizable way. Most companies think their story is boring. It almost never is; it’s just told predictably. The mistake that taught you something, the customer who surprised you, the choice made against all reasonable advice—these are the things a human mentions in word of mouth, and a model can disambiguate them from the rest.
Use AI to accelerate production, never to replace storytelling. If, rereading one of your posts, you feel like anyone could have written it, the problem isn’t the tool, it’s that you’re hiding the part that makes you recognizable.
8. How to understand if and when to change roles
Stay if it keeps you updated. Leave if it leaves you behind.
The most useful framework I’ve read this week. It’s not about money, the title, or boredom. It’s about the direction of the learning curve.
Every role has an implicit learning rate. It’s high at first, then stable, then declining. The tipping point isn’t when you stop learning, it’s when what you learn is no longer useful to you in the world that’s coming.
For founders, the question doesn’t change; the person changes. It’s not you who needs to change the company, it’s the company that needs to change at the same speed as you. If your operational work today is the same as it was 18 months ago, you’re probably slowing down growth, not driving it.
Every quarter ask yourself this question: “What I’m doing this week, am I learning to do better or am I just doing it more?”
9. How to think in two speeds
A classic pitfall in small companies: the founder handles everything and thinks about everything on the same level. He handles issues six months and two years from now with the same sheet of paper, on the same day, in the same frame of mind. As a result, both dimensions are under-optimized.
The correct way to divide the load is not by area, it is by time horizon :
The team solves problems for the next six months
You plan what the organization should be like in two years
If you do the opposite, diving into next quarter’s problems and leaving 2028 to a vision.md file opened twice a year, you are the system’s ultimate bottleneck.
Inside MAKE PROGRESS®, we have a tool to manage this dual track: Now and Next . Now is the pace, Next is the direction. Two different tasks, two different cadences, two different types of questions. It is in this tension that the most powerful OKRs are born.
10. How to be sincere without sounding like an asshole
When you have to say something uncomfortable, your instinct is to be direct, because frankness often sounds like honesty.
But candor without the ability to leverage results isn’t “transparency,” it’s gratuitous pressure . And pressure shuts people down until you hear a “crack.”
The lever is the context in which you say the thing, the rapport built beforehand, the moment you choose, the why that accompanies the what . Without that framework, the same sentence goes from feedback to judgment in five seconds, and the recipient stops listening even while nodding.
Useful honesty leaves room for maneuver. It states the problem, points a direction, and doesn’t close the door. Useless honesty is content to have told the truth, even if no one is in a position to do anything about it.
Three rules of thumb I use in every one-on-one feedback session with my team:
Criticize the idea, never the person
Give feedback when you have energy, never when you are tired.
Before you speak, ask yourself, do I want to fix this situation or do I just want to be right?
These are ten invitations to pause for a moment before returning to work. If you choose one and do something with it this week, for me, this newsletter’s job is done.
Which one made you say , “Here, this is exactly what I’m experiencing right now “? Reply to this email and tell me the number. I’ll read them all, one by one.
ALWAYS MAKE PROGRESS ●↑
Antonio
Antonio Civita, Founder of STRTGY and author of the MAKE PROGRESS® method. I work weekly with founders and teams of Italian companies on growth, operating systems, and strategic decisions. These ten points are based on real-world cases I’ve seen in consulting over the past two weeks. Stay in touch with me on LinkedIn .
Don't miss the next Notes. Every Monday at 7:00 a.m. Free.
Tools and frameworks to unlock innovation in your company and apply Design Thinking, Blue Ocean Strategy, JTBD and OKRs in practice.
My first TEDx, and a phrase that has stuck with me ever since. At Harvard, someone said "fuck AI" and the room erupted. Other leaders were booed for the same ideas, others applauded for the opposite. The 150-year-old economic pact is breaking down, and the confirmation that AI will never correct weak leadership.
A month of silence to accelerate, Qonto and APIs, kids and how hard it is to grow today, augmented leadership, the Bassano del Grappa TedX and the end of the performance, Progress OS, the MCP method and the seventh cohort to become a MAKE PROGRESS® Progress Coach.
Better prompts, agents who make decisions, strategies that lose consistency, branding in the AI era, stop being assholes. Signals to anticipate how the world is changing in terms of how we work, sell, and decide.
Entrepreneurs and managers carry different risks, have different priorities, and often make decisions about different companies. Discover how to transform this gap into a strategic framework that works.
The fastest companies don't do everything faster. They manage two different rhythms. Here are the 5 strategic tensions and the 10 tools to manage them without losing control.
While you update your resume, someone is creating the future. Companies are no longer looking for those who "know how," but those who decide what should be done, starting with the case that shook Silicon Valley in 82 days.
I took the 2030 strategic plan and mapped it out using the MAKE PROGRESS® tools. I'll show you how I did it, piece by piece, so you can apply the same method to your company.
