Share how Developer Community AI helped you for your chance to win

Chinese language support is not good enough

It still has many unanswered questions, so if it were fed with the documentation, I think it could be more powerful. However, my favorite answer so far is this one: https://pt.community.intersystems.com/ask-dc-ai?question_id=120963 it gives a great headstart on questions as that when developing.

I think the DC AI is the most helpful with answering more straightforward questions. For example, on DC every once in a while there's a question on converting dates (for example): https://community.intersystems.com/ask-dc-ai?question_id=129616

Most short questions like this can be answered by AI and if the answer is wrong than it's worth posting a question on a Developer Community.

There are lots of articles in the DC and in the documentation about error handling and I have a few of them bookmarked, but for the most part I always throw $$$GeneralError (5001).  Using this tool and asking about when I should or shouldn't use this helped me learn that there are more buckets to throw custom errors under.  Plus the link to the source article gives even more info on registering custom error codes (I've never seen this article, so glad this tool pointed me to it).  

https://community.intersystems.com/ask-dc-ai?question_id=132835

I got a question from an intern having trouble writing out to a specific file using %Stream.FileCharacter, and thought I'd see how DC AI would do: https://community.intersystems.com/ask-dc-ai?question_id=150746

Not bad! Not quite the way I recommended, so it's just a reminder to give the DC better input. ;)

The DC AI have lots of good posts. I used the https://community.intersystems.com/node/567051 to begin my studies. Good luck to everyone!!

I asked the question: How can i locate the source code line corresponding to an ObjectScript error?

I thought the answer was reasonably good: https://community.intersystems.com/ask-dc-ai?question_id=164044 and would be helpful to many ObjectScript developers.

I tried DC AI and found it quite impressive. Below is one of the questions I asked:
https://community.intersystems.com/ask-dc-ai?question_id=177731
Here are some suggestions for improvement:
Session History: Having a history of asked questions, at least at the session level, would enhance convenience and allow users to track their inquiries more easily.
Reference Links: Instead of displaying links always as "[1]", consider using more readable references like "Ref #1" or "Ref #2". This format would improve readability.

I used it to find detailed steps of how to create my first business process

https://community.intersystems.com/ask-dc-ai?question_id=178450

While it is nice to have, like other have said it still is learning and lacking. The responses weren't as fast as one would expect, but maybe my questions were too complex for it.

I asked how to export global into XML format and the answer was perfectly accurate with the first suggestion: 

https://community.intersystems.com/ask-dc-ai?question_id=51882

Cool feature!

I asked for interoperability adapters for issuing shell calls and the answer was quite objective, although the DC post reference didn't match my question (I guess)... Anyway, here is the link: https://community.intersystems.com/ask-dc-ai?question_id=218088

PS: I tried to google the same question just for fun, prefixing Intersystems to it ("intersystems is there an interoperability adapter for call programs by command line?"), and didn't get any result pointing to EnsLib.Pipe.OutboundAdapter. :)

I just asked "How do I change my password in InterSystems Server Manager in VSCode?"

It gave me three answers: the first one was outdated, the second one was a bad idea, and the third was the right one.

The results can be good, but there's room for improvement.

One of my go-to LLM tests is to provide a zero-shot prompt for ObjectScript code that calculates the distance between two points. As seen in this example (https://community.intersystems.com/ask-dc-ai?question_id=238163), which has some issues.

Most LLMs get close to answering this question, but they often fail with operator precedence and / or the square root function, either misspelling $ZSQR as $SQRT or hallucinating a math library that doesnt even exist, such as ##class(%SYSTEM.Math).Sqrt().

The problem stems from the token volume of other languages outweighing ObjectScript in the gradient descent during training. This causes solutions and functions from other languages to bleed into the ObjectScript code.

RAG powered by IRIS vector search is a good approach to address this, but it looks like it can be improved.

Generally, the LLM often has the right idea of how to answer the question, but not always with the correct details. To use an analogy, the LLM already has millions of recipes, but these might not match the ingredients in the kitchen. If we can tell it what ingredients to use, it will do a far better job at morphing a relevant recipe.

One strategy is to first ask the LLM to break down a question into smaller questions, such as "How do you calculate the square root of a number?" However, the DC AI still struggles with these simple atomic questions about the ObjectScript language:

https://community.intersystems.com/ask-dc-ai?question_id=239065
https://community.intersystems.com/ask-dc-ai?question_id=239114
https://community.intersystems.com/ask-dc-ai?question_id=239125

At a minimum, the vector database should also include these atomic facts about IRIS and ObjectScript.

One approach I've experimented with is to produce a compact document of atomic facts and either include the entire text in the prompt or have the LLM select which facts it thinks it needs first from a keyword list.

This results in a prompt reengineered by the LLM itself:

"InterSystems IRIS ObjectScript Reference: ObjectScript has strict left-to-right operator precedence, use brackets to ensure correct operator precedence, $ZSQR(num) is used to calculate the square root of the given number, $ZPOWER(num,exponent) is used to calculate the value of a number raised to a specified power. Task: How do you calculate the distance between two points."

With this approach, we see that the DC AI gives a much better response:

https://community.intersystems.com/ask-dc-ai?question_id=239152

I'm enthusiastic about this technology's potential. If you need a beta tester for future versions, please reach out. I'm also happy to contribute my ideas further if its of use.

I asked him about SQL-Type KPI. Specifically about getting filters of the entire dashboard inside the KPI. AI says it did not know anything and sent me to ask the community about it.

For simple questions, its working fine. If after the answer I want to refine my question, I need to add to the original question, and not sure if sessions are persistence (like in chatGPT that I can do a "conversation").
when a too long and complex question is being entered, I got:
"This is beyond my current knowledge. Please ask the Developer Community for further assistance."
There is an advantage that some (simple) questions get good referrals to the documentation or community pages

I initially asked about using the “Populate” method in ObjectScript to generate random data and received a useful response. Later, I provided a sample class definition in my query and asked: "how to use populate method for random data". Impressively, it gave a code snippet that successfully generated the data.

Recently, I asked the DC-AI where the streams formatted as %GlobalCharacterStream are saved, as since I suspected that these were responsible for a significant disk usage. 

The chatbot answered: <"Streams formatted as %GlobalCharacterStream are typically stored in the global ^CacheStream by default.">. I found out that I wasn't the only one with this problem and asked further information about the CacheStream global and how to clean it up.

The best solution to my issue was to implement a custom purge task, as I found in one of the sources the bot provided me (that provided purging code too):

1.Cleaning up CacheStream Global

2.Default Purge Does Not Delete Streams

3.A beginners guide to Orphaned data- How as a trust we cleaned up 200+gb

My overall experience with the DC-AI has been quite positive. Provided information is still closely tied to the specific answers given on the forum about the argument and could be more generalized, but I believe we're on the right way.

Links to the discussions below:

https://community.intersystems.com/ask-dc-ai?question_id=249111

https://community.intersystems.com/ask-dc-ai?question_id=249126

Well, it really simplifies the task of search and deep dive into hundred of community articles to get a specific piece of code, in my case, how to export SQL Gateway configurations:

https://community.intersystems.com/ask-dc-ai?question_id=251638