Diffenrent Level of coding with AI

 Important notice, if you want to code with AI at one of the levels, you must master all lower levels, there is no short cut. If you want Level 4 multi agent coding, you must go through Level 0, Level 1, Level2 and Level 3.

Second notice, You only reduce the coding time but software development is a little bit more, maybe around 60-80% more. 

Last remark, all values in the table are good guesses based on my experiences. 

 So, now you have to find your sweet spot.

Level	AI usage	Source code control	AI cost	AI Gain
0	no	full	no	No AI slope debugging
1	auto completion	full	no	minimal
2	prompt	full to most	non to minimal	significant reduces search time
3	agent coding	less to non	yes	coding time significant reduced
4	multiple agents	no	significant	coding time reduced to the half

 

AI assisted software development

Overall: now in 2026 software development works pretty well under some conditions:

1.  High quality code base
2.  High experienced software developers

This is the perfect combination for successful AI assisted software development. AI is getting faster and tools getting better, it's a lot of fun to code with AI. At the moment AI is in the storming phase.

You want an example, here it is a small Golf trainings app with mostly complete AI generated frontend and AI maintained and extended backend. 

 

http://golf.ebert-p.com/

Used AI:

1. aider, ollama, qwen-coder (different versions) - running local models
2. grok build beta 

 

Bullshit Asymmetry in Software Development or Why is Software recunstructing so expensive

Bullshit Asymmetry in Software Development means that is very easy to develop a new system on a green field. You can easily go on putting new features into the code  but over time, often faster then slower, you have to put more and more effort to add new features until you reached the culminating point. At the software development culminating point your resources (developers) are exhausted, no new feature can be implemented in acceptable time. Next the development team is wearing out and then I'm appearing.

1.  Software reconstructing is more expensive then writing the software. The costs are asymmetric. Build a new software is much cheaper then reconstruction/rescuing legacy software. 
2. But software reconstructing lowers the risk of a complete miss. The improvements are coming at a constant speed. No new teams are needed. Reconstructing software can be realized by the origin team, maybe extended by some external experts. 

 There is a third way, re-implementation of an existing system. This way combines low risks, like software reconstruction with the low costs of building a new software.

  

	Write new software	Reconstructing software	Re-implementation
cost	low	high	low
risk	high	low	low
results	after long time	immediately, slow flow of improvements	after not so long time
team	new team needed + one insider	existing team + some external experts	Only some seniors needed
new features	are implemented	slow but steady, accelerating	no, but faster after release

Easy Ways to increase efficiency in software companies: Meetings

 

The efficiency in many software companies are low. You may know it, endless meetings with to much people never needed. Presence counts more then results. But here some simple things that increasing the efficiency are really easy to implement:

1. One meeting free day a week (Meeting-frei Freitag)
2. No back to back meetings, at least 5 min break between
3. Time boxed meetings
4. Regular breaks in long meetings 

This ideas will not fix every thing but they are simple and easy to implement. And these ideas could be a turning point to more focused, result focused work. 

 

Fun fact, after introducing a meeting free Friday, after some time, a lot of meetings appears on the meeting free Friday, because there was a lot of free time. But now the Thursday is now meeting free. 

AI, LLM create vulnerable software

First: AI, LLM are a very powerful tool for software developers. 

The following lines will indicate a limit of LLM in software development. 

So you aware of it and can handle it right. 

I found this problem by incident. LLM like qwen3-coder:30b generating code that uses outdated libs and frameworks containing CVE's. 

 

Conclusions 

1.  LLM's are not perfect writing code but don't drop it.
2. Many (Most, All) LLM's living in the past.
3.  LLM's  generate projects with outdated vulnerable dependencies. Be aware of it.  This will time problem will affect all ideas the LMM is generating.

Test

 The task: create a Spring Boot web app with qwen3-coder:30b (ollama, aider) 

1. pom.xml contains Spring Boot Starter Parent 2.7.0, uh that's a little bit outdated
2. Release date was May 19, 2022
3. contains at least 10 CVEs
4. Select Java 11 

Ok, I give a helping hand 

1. Task: create a spring boot web app, spring boot 4 
2. pom.xml contains Spring Boot Starter Parent 3.2.0, uh that's a little bit less outdated
3. Release date was November 23, 2023
4. contains at least 2 CVEs
5. Select Java 17

Looks that the model is living in the past, at least 2 years. So I give are more modern model a try: qwen3.6:27b 

1. Task: create a spring boot web app, spring boot 4  
2. pom.xml contains Spring Boot Starter Parent 3.3.0, best I can get but still far away, gives me a hint to 4.0.0-M1 
3. Release date was May 23, 2024 
4. contains at least 7 CVEs 
5. Select Java 17 

 

Expect the Unexpected: Some Words about Errors

     For all ‘Happy Path’ developers, proper error handling is a fundamental feature of software, and it is quite possible that error handling is for half of your code.

 

Expect the unexpected: not all English postboxes are red. This Victorian one stands in Dublin.

 

There two types of errors:

• Retryable Error
• Permanate Error

Retryable error: a temporary issue has occurred. a retry is likely successful. Example causes: firewall reboot, db maintenance, HTTP 5XX.

Permanent error: error where a retry will always end-up with an error. Example: HTTP4XX, NPEs (Null Pointer Exceptions).

After a retry sequence, Retryable errors should become permanent errors.

And at last, don’t forget Logging and Monitoring of errors.