Journal
Journal: method
On this topic, 60 articles from the journal, from practical guides to behind the scenes of the product.
User stories explained simply: how to divide a product into steps
You may have been told your project will be divided into 'user stories'. The term belongs to developers and hides an idea you already practise when commissioning work: describe a step by what can be verified at the end.
Read the article
How to write an application brief: the perfect brief does not exist, start messy
A brief's blank page blocks more projects than the budget. A disorganised twenty-minute draft is enough to start, provided someone then asks the questions you missed and returns a clean document for your signature.
Read the article
Zero clients in six months: why SMEs reject custom development
A provider contacted around 1,500 companies in six months and got four meetings, all lost when price and timeline entered the conversation. SMEs' refusal is explained by two numbers, and one just changed.
Read the article
Documenting software's business rules: extract them before it is too late
The value in your software is the years spent deciding how things work in your business. Those decisions live in two heads and code nobody rereads. Put them in French, and understand why the document outlives the code.
Read the article
Software migration without interruption: rebuilding in parallel, week by week
An uninterrupted software migration is not decided at six on a Monday morning. It unfolds over weeks, keeping the old tool alive while the new one is built alongside it, with a period maintaining both. Here is the week-by-week plan.
Read the article
Behind an application dashboard: showing that a product is running
Our second tester manually moved four steps on their board because the application no longer did it. On August 26, we rebuilt how Maestro shows its work: ten workstreams, 43 defects fixed, and a lesson for every dashboard.
Read the article
Everything is green, yet it is wrong: how to check what AI built
An advertising professional who cannot code built an application in a week with more than seven hundred passing tests. The import had skipped one file in three, and the tool proposed deleting orphaned data. What green proves, and what you need to inspect yourself.
Read the article
Mobile application specification: the eight decisions it must contain
Online specification templates ask for your email and return an empty document. The eight decisions below are what a provider or agent team will ask anyway, and nobody can make them for you.
Read the article
Project scoping: an AI that asks questions is better than one that guesses
A founder sums up the problem in one sentence: without a real specification, the tool designs things its own way and they spend their time catching up. Scoping through questions solves this before the first line of code, provided someone asks them.
Read the article
Build an app with Claude Code without knowing how to code
Subscription, usage limits, interfaces without a terminal, a brief, five steps and checks: how to have Claude Code build an app when you can't code. Prices and documentation checked on October 10, 2026.
Read the article
Keila, an iPhone app built with Maestro
Finding a kosher, vegan or vegetarian restaurant in Paris without calling three people: that's the idea behind Keila. Both the app and its website were built with Maestro. What it does, how it was made, and what's left before the App Store.
Read the article
Maintaining an AI-built application: understanding what you had built
Vibe coding works until someone has to take over the application. Two accounts, one from a founder and one from a French developer, describe the same wall. What brings it down is a document you have read and approved.
Read the article
When AI reads my company's documents: accuracy from 26.7% to 86%
An AI given your contracts answers confidently and gets things wrong half the time when the document is long. Mistral measured the gap between guessing and searching. That figure also explains why scoping starts with your files.
Read the article
AI assistant outage, project blocked: August 16 and what stays with you
On August 16, 2026, Claude Code was unavailable for forty-two minutes. For anyone having software built by an agent team, the question that follows matters more than the incident: depending on where your documents and code live, a provider outage takes away an evening or your whole project.
Read the article
Rolling back your application without a developer: your product's versions
A change goes wrong and the application starts limping. Without a developer behind you, two guarantees are enough for peace of mind: see what changed, and restore yesterday's product with one action. How it works, and where it breaks.
Read the article
Behind an application update: a signed-out assistant no longer goes unnoticed
A tester tells us they installed their AI assistant without ever signing in. The software treated 'please sign in' as completed work. Here are the three fixes released in version 0.1.17, and the two defects we are leaving open until the next pass.
Read the article
The 2009 Access file is still running: replacing an Access database gradually
The Access database an intern wrote in 2009 has run your orders for fifteen years, and nobody dares open it anymore. Replacing it all at once is the best way to lose what it knows. Here is an exit plan that keeps it alive during the transition.
Read the article
Check your application before launch: the 23 vulnerabilities an audit found
The final week goes into announcements, screenshots and first-customer notices. Two pre-launch security reviews, described by a commissioner and a provider, show what else the week should contain and why the list fits into five ordinary working days.
Read the article
Build an internal application for a small team
Assign decisions, handle conflicting reservations and run a pilot: a practical guide to building a shared tool with the people who will use it.
Read the article
Maestro beta: request access and prepare your first project
How to request an invitation, prepare your AI access and choose a useful first trial. A guide to exploring Maestro on Mac with a clear goal and realistic expectations.
Read the article
Choosing an AI assistant and model in Maestro
Understand assistants, models and agents, compare two options on a small task, and prepare a switch without losing your project decisions.
Read the article
Subscriptions, credits and tokens: understanding AI usage
Why do your assistant and provider show different figures? Learn to read the units, check calculations and connect the numbers to a working session.
Read the article
Build an application: choosing between local and cloud tools
Separate four places: construction, project files, AI processing and the finished application. Compare access, costs and recovery through a practical trial.
Read the articleValidate your app idea with Margaux before building
Between an appealing idea and a confirmed problem, there is often some research to do. Prepare your observations, work on the brief with Margaux and decide what to check before building.
Read the article
Human approval of AI-generated code: GitHub opens the door, off by default
Since September 1, GitHub's assistant can approve code changes rather than merely comment. The feature arrives switched off, adjustable per project, and that default says more than a speech about human oversight.
Read the article
Why software projects fail despite requirements: silent disagreement
The document was complete, reviewed and approved by everyone. Nine months later the product launched and nobody wanted it. What these stories reveal about obtained versus real agreement, and what an approval gate must provoke.
Read the article
The quote is not the cost: application development budget overruns
An application budget overrun almost never comes from the hourly rate. It comes from what nobody wrote down. To the customer, the bill grows without reason; to the provider, thirty days sold become forty-five. Both describe the same thing.
Read the article
Artificial intelligence adoption: 18% of French businesses, and who is missing
INSEE measures a tripling of artificial intelligence adoption by French businesses in two years. The overall figure hides a sharp size gap, a barrier other than price and an entire category the survey does not even examine.
Read the article
How do you describe your app idea to AI?
You know what your app should do, but not how to ask for it. Start with this reusable example, then answer a few questions to make your idea clearer without technical vocabulary.
Read the article
Behind the scenes: the bell that does not shout, rebuilding an application’s notifications
Our first useful notification arrived drowned in everything requiring nothing. Five comments in one session, one simple cause, and a rebuild delivered the same day: the story of an avalanche we created ourselves.
Read the article
Read application requirements without a technical background
Take an example from your work and follow it through the document. Six questions help identify what is missing, what is unclear and what you can already approve.
Read the article
AI fixes one thing and breaks another: what should you do?
You add a phone field and the confirmation email disappears. How to describe the problem, limit the next fix and check what needs to stay unchanged.
Read the article
Your first application looks finished, but launch is stuck
The screens are there. Yet a forgotten password or an empty list can stop a user. Here is how to turn those “last details” into a clear work list.
Read the article
Automated decisions and GDPR: €825 million for missing human approval
An automated decision depriving a driver of their account without human review: the Dutch authority, working with the CNIL, valued the failure at €824,990,000. Human approval is covered by an article of GDPR, and its absence has a price.
Read the article
Your developer has stopped replying: where should you start?
The site still works, but you do not know who can change it. An access list, a copy of your information and a review of its condition help you prepare the next steps.
Read the article
Build an app without coding: a guide to your first project
From a need to the first real use: choose an approach, write a brief, plan costs and check the result. A complete example to help you move forward without technical vocabulary.
Read the article
Application development is conducted like an orchestra
A score written before playing, sections with clear remits, a conductor holding the baton without playing an instrument: the metaphor structuring Maestro, embraced all the way to its logo.
Read the article
AI makes experts slower: why approval must be an architecture, not a chore
The METR study measured experienced developers as 19% slower with AI, while they believed they were 20% faster. Review and verification consumed the gain. The problem is not review: it is where we put it.
Read the article
Self-service skills: the supply chain nobody is watching
The github/spec-kit repository shows 130,054 stars a year after launch. The AI agent skills ecosystem is growing just as fast, and installing a skill on demand is becoming the new curl piped into a terminal.
Read the article
Prototype, MVP or application: when is it ready for real users?
A prototype explores an idea; an MVP tests a limited use. Before involving real users, check the complete journey and record the evidence needed to approve a pilot.
Read the article
Can you keep your application if you change tools?
Your customers and their appointments are already in the application. Before depending on a tool, check what you can retrieve and what you would need to continue elsewhere.
Read the article
A first requirements draft in one hour, with an example
Set aside an hour for a first draft: the problem, two journeys, rules to decide and what can wait. A workshop to revisit, not a document to freeze.
Read the article
How do you check an application when you cannot code?
Two customers book the same time: what should happen? This example explains what a test is for, why to prepare it early and what you can check yourself.
Read the article
Independent professionals: which small tool could help your practice?
Your main software works, but a separate spreadsheet is still essential. That file may be the starting point for a more useful tool, provided you choose a specific need and keep your professional requirements.
Read the article
Before approving an AI proposal, what should you check?
A document can sound convincing and miss an important rule. Three things to check so your approval really matches what you need.
Read the article
When the score falls: what our software comparisons showed
In August 2026, a newer version received a lower average score. We compared the software outputs in pairs to understand the result. Here is what that check showed, without converting the verdicts into a new score.
Read the article
Why we also describe failures in our AI agent tests
A good score can hide a missing feature. Two results from our internal trials show why we need to examine errors and specify what we actually measured.
Read the article
Can a team of AI agents deliver software? Our tests and their limits
Our tests covered an initial development batch. Here is what they showed, how we evaluated it and what still needed checking before real use.
Read the article
Why Maestro moved away from BMAD: our decision and test results
We replaced BMAD with our own method to organise Maestro’s steps more directly. A campaign of nine scenarios in August 2026 informed that decision without establishing a general ranking of tools.
Read the article
Vibe coding, explained with a small application
You describe a need, AI writes code and you try the result. A customer-tracking example explains what this approach offers and what you need to check.
Read the article
“Waterfall strikes back”: our response, point by point
225 points and 191 comments: the Hacker News thread accusing spec-driven development of reviving waterfall made an impression. Three precise criticisms, a quantified response, with no evasion.
Read the article
Business software rebuild: planning data migration and the switch
When should you rebuild a business application? Understand the options, check the warning signs, preserve your data and test the change before switching.
Read the article
How an AI agent helps us test Maestro’s responses
A user may reply too quickly, request a correction or reverse a decision. Our test setup simulates these situations to look for blockages, within clearly defined limits.
Read the articleFirst week of beta: what broke, what we fixed
Our first tester never saw the second screen. An honest account of the first days: three real defects, what they taught us, and what remains open.
Read the article
Six days of testing: what we measured against BMAD
From August 5 to 10, 2026, we compared versions of our method with BMAD and then with each other. The results concern documents and an initial set of features, not complete applications delivered to customers.
Read the article
Your application’s rules are worth keeping too
A discount, a cancellation, restricted access: these decisions explain your application. Keeping them in writing makes corrections and handovers easier.
Read the article
Application requirements: a guide, example and free template
A booking example to describe who uses your application, what it should do and how to check the result. Includes a free text template in French to adapt.
Read the article
Directing AI agents without coding: the decisions you make
To build a calendar or a tracking tool, you already know an essential part of the job: the need. Here is how to explain it, follow the work and check the result.
Read the article
The deliberate opposite of a terminal
Why Maestro looks like a concert hall rather than a black screen: the setting is part of the product.
Read the article
Why nine agents have first names
Margaux, Victor, Salomé, Maurice, Constance, Marcel, Amélie, Félix and Léa. Names that make the work easy to follow.
Read the article