Is anyone using an AI product for coding with any success?

I ask this question after two recent incidents:

(1) I was searching google for answers to an issue compiling an older, abandoned FOSS C project... I am not fluent in C... and wound up
interacting with Gemini. It got me on the right track in the sense that we fixed the compile error (caused by compiler and library updates) and got it running.
(2) reading about and using Claude's BBS, which was apparently written with
the assistance of AI.

In my personal case, I found that Gemini was good up to a point, but had trouble remembering what we were working on after a while -- after we got
into the weeds about squashing some memory leaks. We got sidetracked a little and then it completely forgot what we were doing. It also hallucinated
some.

However, after interacting with Claude's BBS and seeing what it looks like,
I have a very old C project (DOS!) that I wouldn't mind running by some
other AI product to see if it could help me fix a nagging bug. I also have a few other ideas I wouldn't mind trying out.

So I am wondering if anyone has had success with AI, which AI, free or
paid, and maybe if it is some AI that you are hosting yourself (and, if so,
is it linux based?)?

Thanks!


* SLMR 2.1a * The bold print giveth and the fine print taketh away.
--- SBBSecho 3.28-Linux
* Origin: Capitol City Online (1337:3/103)

Is anyone using an AI product for coding with any success?

Yes and no. I find that I have to steer any agent in the right direction. Always. Matter of fact I am happy about that ^^. The amount of times any AI agent goes completely bonkers and out of style of the project is crazy.

Even with short focussed tasks it needs nudging and guide. Vibe coders must produce absolute slob :P.

In my personal case, I found that Gemini was good up to a point, but had trouble remembering what we were working on after a while -- after we got

Gemini was the worst experience for me. Claude Opus is doing really well lately.

What helps a lot is a custom prompt with important styles imho.

I was using AI to expand on a Flask project which heavily relied on abstraction and polymorphism and it could not get its head around the full scope of the project. Ended up going through the source myself.

oP!

... Are those cookies made with real Girl Scouts?

--- Mystic BBS v1.12 A49 2024/05/29 (Linux/64)
* Origin: TheForze - bbs.theforze.eu:23 (1337:1/109)

Is anyone using an AI product for coding with any success?

--previous text--

That said, AI absolutely helps me with a enormous time boost on many small tasks. A lot of custom Wordpress plugins and snippets which I made in the past 6 years definately get improved a lot by passing them through Claude Opus.

Also working on a lot of SQL queries for Sap Business One which I am improving with AI. Few years ago I could spend half a day on creating a SQL query to get some KPIs fleshed out. Now 10 mins to a hour on same scope.

I think for smaller scripts AI is amazing. On large projects I use it for insights and implement meticulously described changes to avoid it going mental.

oP!

... There is one God, but which one is He?

--- Mystic BBS v1.12 A49 2024/05/29 (Linux/64)
* Origin: TheForze - bbs.theforze.eu:23 (1337:1/109)

Yes and no. I find that I have to steer any agent in the right direction. Always. Matter of fact I am happy about that ^^. The amount of times any AI agent goes completely bonkers and out of style of the project is crazy.

Good to know.

In my personal case, I found that Gemini was good up to a point, but had trouble remembering what we were working on after a while -- after we got

Gemini was the worst experience for me. Claude Opus is doing really well lately.

What helps a lot is a custom prompt with important styles imho.

Someone else mentioned Claude as a better source.

I was using AI to expand on a Flask project which heavily relied on abstractio
and polymorphism and it could not get its head around the full scope of the project. Ended up going through the source myself.

Thanks for the input! I am not a C programmer (COBOL!) so going through some old C code could be a challenge. I am already prepared to do some
searching over it and seeing if I can fix it without assistance first.


* SLMR 2.1a * You're so vain / I bet you think this tagline's about you
--- SBBSecho 3.28-Linux
* Origin: Capitol City Online (1337:3/103)

Re: Using AI for coding
By: Dumas Walker to ALL on Fri Feb 20 2026 10:17 am

Howdy,

So I am wondering if anyone has had success with AI, which AI, free or
paid, and maybe if it is some AI that you are hosting yourself (and, if so, is it linux based?)?

So over the years, I've been experementing with AI just to see how "good" it is.

I started with ChatGPT (year or two ago), and asked it to write an erasure coding class for PHP - it spat out function placeholders, and I asked it to complete each placeholder (several times, as it had several of them) and the final code didnt work.

When I gave it the errors, it gave fixes that didnt work. I put it down to being useless.

I tried it with a couple of other models, cant remember which ones, and it to didnt give me anything that was useful at all.

Then late last year, I tried claude, and give it the same instructions. I was super impressed, not only did it give me working code, I had a discussion with it over parity with the data, or in seperate blocks, and forward error correction over normal parity loss. I asked it questions like why one method over the other, what is used the most, benefits and cons of each, etc.

I thought I was talking to an expert and I had code that could do all different implementations of parity and error correction.

Just recently, I've been using good old google search (gemini?) to build regex expression (to actually pull apart type 2 packed messages in FTN packets). I was impressed that I never told gemini that it was fidonet but when I asked it to create a regex to pull out the kludges and gave it a data example, it not only knew I was pulling apart fidonet packets, it gave me some other suggestion for when data was in a different layout (eg: sometimes no origin line).

For coding, I think AI has come a long way, and where it I have found it useful, is not only has it provided working code, it provides a breakdown of what element does and why it works, pros/cons and considerations. I can see it taking away the bulk of the work when programming.

...ëîåï
--- SBBSecho 3.29-Linux
* Origin: I'm playing with ANSI+videotex - wanna play too? (1337:2/101)