Proposal 2024: GPT-Based Chatbot Integration for Drupal

Hi @ubulinux,
I am maintainer of AI chat user interface → , AI prompt engineering → and others
I am not sure how Google Summer of Code works, but I will be able to help out. If not as a mentor, so as maintainer.
Project hours listed here is quite low, because AI chat is a large feature with lots of potential.
Since a lot of work is already done (way more than 150), maybe its good idea to work on improvements. Recently I added a lot new features to-do to the issue queue https://www.drupal.org/project/issues/aiprompt?status=All&categories=All →

@minduagasd → , Thank you for reaching out and offering your assistance! I'm glad to hear that you're interested in contributing to the project. Your expertise in AI chat user interface and prompt engineering would certainly be valuable, whether as a mentor or maintainer. Your suggestion to increase the project hours to 350 aligns better with the scope of work and the requirements of Google Summer of Code. I appreciate your attention to detail in this regard. I'm definitely open to working with you both as a mentor and collaborator. Your experience in the field brings a lot to the table, and I believe we can work on this issue together.

@mindaugasd → , Thank you for reaching out and offering your assistance! I'm glad to hear that you're interested in contributing to the project. Your expertise in AI chat user interface and prompt engineering would certainly be valuable, whether as a mentor or maintainer. Your suggestion to increase the project hours to 350 aligns better with the scope of work and the requirements of Google Summer of Code. I appreciate your attention to detail in this regard.
I'm definitely open to working with you both as a mentor and collaborator. Your experience in the field brings a lot to the table, and I believe we can complete this issue together.

Created another example of very useful feature to work on 📌 Chat in the sidebar (drupal off-canvas support) Active

To learn more about google summer of code, I read all comments and articles (1,2,3,4,5,6) on very similar project last year:
#3339015: Proposal 2023: Chatbot interface for Drupal documentation and contributor onboarding →

But:
1) Many comment applications and articles before were written with chatgpt.
2) Final project result is quite basic

If this project would be done this year too, then it has to aim for next level quality.

Goal can be solving real tasks on modules:
1) AI chat user interface https://www.drupal.org/project/aichat →
2) AI prompt engineering https://www.drupal.org/project/aiprompt →
3) AI personal assistant builder https://www.drupal.org/project/aipersona →
4) AI developer assistant https://www.drupal.org/project/aidev →
5) AI work https://www.drupal.org/project/aiwork → or AI interpolator integration with chat https://www.drupal.org/project/ai_interpolator/ →

But also sending links to other big Drupal AI ideas, which could be done in combination with chatbot project:
1) 📌 Use ChatGPT for solving Drupal issues to increase rate of development Active
2) 📌 Training a Large Language Model for Code (Code LLM) - an expert Drupal A.I. Active
3) 🌱 Fully fledged website and module development with Artificial Intelligence Active
4) 📌 [META] Drupal could be great for building AI tools (like ChatGPT) Active
5) ✨ Talking with content - is it the content management of the future? (plus Python side question) Active

I reduced time back to more realistic 175 hours, but this can be increased in case applicant is very hard working and determined to achieve a great deal.

Will post another update as I will continue to explore Google summer of code.

For information of gsoc administrators:

@ubulinux have contacted me and we talked about his project, as well as, what is Drupal, what is drupal.org, how to try Drupal and similar. @ubulinux is motivated and would like to find the mentor for this project.
As for me, I won't take a path to be a mentor, but this remains true:

I will be able to help out [...] as maintainer.

In the case projects I maintain will be useful for @ubulinux to realize his project.

@mindaugasd I been following this chat thread, i found the conversations very interesting. what are your thoughts on the primary challenges we might face in terms of scalability and data privacy? I mean I do see this integration will enhance the Drupal community's engagement and the overall user experience.

@stanzin

what are your thoughts on the primary challenges we might face in terms of scalability and data privacy? I mean I do see this integration will enhance the Drupal community's engagement and the overall user experience.

if talking about updated description of this project, then these features described are usually outside the scope of Drupal, and are part of analytics suites like Google Analytics → or Matomo Analytics.

So first of all, this project needs a proof why it is useful to do it on the Drupal side, as there are specialized services people already prefer to use. Maybe these dedicated services already have a chatbot, and if they don't have a chatbot yet, maybe it is for a reason (too complex, not useful enough or work in progress).

But there is a way to solve this. Data could remain outside of Drupal and users can continue to use a regular service like Matomo. Since data could be queried from external matomo service in a similar fashion like this module does https://www.drupal.org/project/Matomo_reports → and then use https://www.drupal.org/project/aichat → and https://www.drupal.org/project/aiprompt → to talk with AI about this data. In that case, "scalability and data privacy" would be the same like usual - data would be outside of Drupal and all operate under regular website privacy policy.

Second proof I would like to learn - why analytics chatbot itself is useful:

• Who will ask AI analytics questions?
• What those questions will be?
• What will be achieved by asking AI those questions? AI answers will be about what?

Third proof I would like to learn, why not wait until chatbot is done on Matomo or Google side, and use it there. What is the value added having a chatbot on the Drupal side (while querying data from Matomo). Why that is better than going directly in their interface?

Thanks for the detailed breakdown, @mindaugasd! I gotta say that I'm not deeply versed in AI chat or analytics platforms like Google Analytics or Matomo, but your points make a lot of sense. I agree that enhancing Drupal's integration with them, instead of duplicating, could be more efficient and address scalability and data privacy more effectively. Which I didn't think before ;)

However, even with my limited expertise, I can see that a well-integrated chatbot might significantly improve user engagement and streamline administrative tasks within Drupal, such integration could be a big win for a lot of folks including me.

Bdw now I am curious. Any chance you have examples of any platforms have successfully integrated similar technologies, or specific features you think we should consider when designing our chatbot?

Any chance you have examples of any platforms have successfully integrated similar technologies, or specific features you think we should consider when designing this chatbot?

Question is too broad, please specify. Especially because we have a lot of unanswered questions I asked before. And now, also, I would like to learn what you have in mind by "improve user engagement and streamline administrative tasks"?

Hi @mindaugasd, I've proposed a Google Summer of Code project this year focused on AI-driven SEO automation for Drupal. Given your deep expertise in AI integrations and community projects, your involvement as a maintainer could be incredibly impactful.

I have secured a mentor for the project, and I understand you're quite busy, but having your support would be invaluable, especially if the project progresses well. Here's the project link : https://www.drupal.org/project/gsoc/issues/3425856 →

I've also discussed potential integration with the SEO AI team, led by Adrian, as they currently have a sandbox project that could benefit from this module integration, serving this module as a base for future innovations. I look forward to your thoughts! https://www.drupal.org/project/seo_ai →

thanks for your constructive feedback. I want to clarify my approach regarding the chatbot functionalities, which might seem to shift from the initial analytical focus to a more interaction driven approach.

direct user interaction, the chatbot to respond immediately to user queries, similar to how Slackbot operates, this reduces user frustration and improves satisfaction. This real-time interaction can significantly enhance the user experience by providing instant support and answers, which i think is so important for user retention and satisfaction.

current rrupal documentation is brutal, it is quite dense and intimidating for new users. having interactive tutorials, we can help both new and existing users navigate and maximize Drupal’s functionalities. This approach not only makes the platform more accessible and user-friendly but also encourages greater user retention and active participation

automating admin tasks, streamlining managing user permissions and content updates. by automating these tasks, we not only enhance site maintenance efficiency but also free up administrators to focus on strategic growth and improvement initiatives.

I was thinking more in terms of refining the chatbot's functionalities to extend the original analytical aspirations by providing tangible, immediate benefits to both users and administrators. I got carried away.

I look forward to your thoughts )

Hi @Stanzin,
I will look about your another SEO project and let you know my thoughts.

shift from the initial analytical focus

I think moving away from analytics to user helper chatbot is a good idea, it makes total sense and I fully approve it.

I would like to know more about how you are working on both SEO and helper chatbot projects at the same time? :-)

Keep in mind, I don't know the details how Google summer of code works. Will you be working on both projects, or will you still select one of them, or?

No actually I am just contributing to seo ai. This project was proposed after the April 2nd deadline for proposal submission so I was not aware of this project. I saw your conversation and you pointed out so many resources couldn't help it but check it out, I was just curious about your idea ! I learnt so much from this conversation. SO COOL ! For this year I have only applied for SEO ai. l thought your insights in ai and prompt engineering, and as a senior engineer will be highly valuable for the seo ai project )

Imm, interesting, so what do we do with this issue?
On the other hand, on drupal.org issue was posted "Tuesday, 2 April 2024, 17:09:00 UTC" while deadline was "April 2 - 18:00 UTC", so it is one hour before the deadline, or not?

That is great that resource have been helpful to you :-)

Don't take my word for it, but I think as long as the proposal for this project is submitted on the official google summer of code portal it should be fine. That is @ubulinux task as a contributor to submit a proposal so the org admins can rank the projects and select them, as for the issue can be submitted thereafter, the point of having an issue queue is for the students and other mentors to browse through the idea list proposed for the summer project. You can entirely not submit a issue queue, and turn in the proposal that should do as well !

@Stanzin, I submitted the proposal before April 2nd on GSoC's portal. I was just a little late while submitting the issue on Drupal, but I did that too, even before the actual submission time.

Hey @ubulinux that checks out, awesome you are good then. I was a little worried there )