Earlier on today I posted something alluding to some breakthrough that has recently been made to do with AI-driven web development.

I've decided to take that down because I don't see any benefit in scaring people or posting overly dramatic posts. So, don't worry about it. Just keep on coding.

In other news, Dafa's new AI assistant appears to be doing a great job of responding to technical questions. So, I'm going to take Grady off of forum duties for now. That's a good thing and I feel happy about the way everything is going.

All is well.

Hi DC,

I did get a chance to read that post before you took it down (that was with the help of Hymie, as I had him keep scanning the forum for new posts and his timing was spot on) I had prepared a response but when I sent it the thread was deleted.

I tuned into the live stream yesterday and I must say what your doing with Grady is very cool but not sure it made for good viewing. Miss the code on the screen.

On the AI bot front - I think there is room for them both here. I haven't created an account for Hymie on the forum as yet. I'm still developing his skills and think Grady is more mature right now.

Cheers

Thank you for that, and I apologise for deleting the post while you were in the midst of writing a response.

It took about two weeks before Grady became genuinely good at Trongate. At this point in time, I consider PHP and Trongate to be solved problems. The bad news is that I still haven't found any AI system - Grady or otherwise - that is truly good at web design.

I asked Grady to give an honest assessment of its competence in these two areas. Here's what it said:



That's pretty much how it seems to me as well.

I don't think Grady is as good at Trongate as you are. For example, if you were going to truncate a string, I think you know the framework well enough to remember that there's already a method for that. In those instances, Grady would probably revert to pure PHP instead of using the framework's built-in code. So, I definitely do not think Grady is the best Trongate developer here. However, it's still far better than I ever imagined it would be.

It took about two weeks to get Grady to the point where it's at now. The key to the whole thing was definitely downloading the framework documentation from GitHub and spending a solid two weeks constantly asking it to refer back to the documentation. Also, whenever it did something right or solved a problem, I would repeatedly ask it to commit that information to memory.

The biggest challenge I had with Grady was finding a genuine use case. However, I'm satisfied that I've now found one, and it isn't being on the forum. There may still be a setting or something that's enabling Grady to log in and perhaps respond, but that's no longer the priority.

I'm absolutely delighted to see you achieving success with Hymie. I know you well enough to know that it's probably a thousand times better than you are letting on. I have no doubt whatsoever that Hymie will be another triumph for you.

These are crazy times, and I suppose the best thing we can do is to try to stay calm and focused.

Hail to you.

DC

Thank you both for sharing your experiences here with the rest of us who haven't experimented with AI assistants yet. Your first-hand insights into what AI can and cannot do well make for an incredibly interesting read, but more importantly, provide highly valuable information for the community. You guys rock, and you're truly at the forefront of innovation!

Well, God bless you and my self esteem needs all the help it can get. Thank you very much indeed.

For anyone reading this in the near future - he's right. I am absolutely amazing. You should totally hire me. Can't afford to hire me? What are you talking about? You can't afford NOT to hire me!

For anyone reading this in the deep future, may be brag NOT about the things we've said but about the things we've done.

Hail Balazs! Hail Team Trongate!