Why does AI agree with everything I build?
It is built to get you what you asked for. It rarely tells you the idea, the design or the code is wrong unless you ask it to look for problems, and even then it tends to find a way to agree
Blog
AI is a yes man: the three things it leaves out of your app
AI builds what you ask for and agrees with you the whole way. Observability, scalability and security are what it leaves to you, and why an engineer should check them
Free, no call
Get every Fix AI Slop Code episode
The code it wrote, the code it should have written and a check, for every episode
The short answer
AI is a yes man
AI builds what you ask for and agrees with you while it does it. It does not push back, and it leaves out the three things an experienced engineer checks first: observability, so you can see what broke; scalability, so the app holds up when people use it; and security, so their data stays theirs.
What is observability?
Being able to see what your app is doing once it is live: logs, errors and alerts. Without it, you find out something broke when a user tells you
When should I bring in an engineer?
Before people depend on the app. That is the point where a missing rate limit, an open database rule or a silent error starts costing you users and money
Watch the video
Watch why it matters
The same idea in a short video. It plays here, and nothing loads until you press play
Everything feels right while you build it
When you build with AI, every step feels like progress, because the model agrees with you. That is the trap. It agrees because that is how it is built, and that tells you nothing about whether the idea or the code is good. It will get you what you want for your idea. It will never replicate what an engineer with experience would have caught along the way.
The three things it leaves out
- Observability. Can you see what your app is doing right now? Logs you can search, errors that reach you, an alert when something stops working. AI builds the feature and skips the part that tells you the feature broke.
- Scalability. Will it hold up when real people use it at the same time? The code that works for one tester often falls over at a hundred, or charges twice when two requests land together. Our note on retries that act twice shows one way it happens.
- Security. Can one user see another user's data? Can a script hammer your sign-up form? Two of the most common answers are in database rules that are on but let everyone in and rate limits that do not limit.
You can train your AI to check for these, and that is where things are heading. Today, a person still has to validate what it creates.
The development loop
If you do everything yourself with AI, it is easy to get stuck building, building and building, and never ship, because there is always one more thing that does not quite work. An engineer's job is to get the product working the way you think it should, for you and for your users, so you stop wondering why it breaks.
What to do about it
Have someone technical look at it before people depend on it. That is what an App Check is: an engineer reads the app AI built and tells you what will break, in order of what it would cost you.
Free, no call
Get every Fix AI Slop Code episode
The code it wrote, the code it should have written and a check, for every episode
Keep readingEvery post
The coding agent never holds production credentials
An agent with a production database URL will sooner or later run a migration or a cleanup against it, so production secrets never enter its env
Keys go in headers, never in URLs
A URL is written to server logs, CDN logs, browser history and error trackers, so an API key in a query string is a key in five places
Backups the app cannot reach, and one restore you have actually done
A backup on the same account the app or an agent can delete from is not a backup, and a restore you have never run is a number you do not have
Bring what you have. We’ll tell you where it stands
You’ll leave the call knowing what’s risky and what to do next
- 30 minutes
- Free
- No slides, no sales team