There are three options for selecting and tailoring a Gen AI model for your organization:
1- Use a public-purpose model, like Chat GPT and the likes
2- Purchase a specific SaaS enterprise AI solution
3- Use or customize a cloud-based secure LLM
These differ in several aspects, among which are cost and privacy – being least in option 1 and most in option 3.
While selecting option 1 is very straight forward, did any of you select and implemented options 2 or 3?
How did you pick which option?
Where did you procure the solution?
Can you share some of the challenges you faced?
How much time did it take to have the solution ready and running? Saving Changes...
Sort By:
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
First of all sorry if I did not understand well your post. Please let me say my recommendation is searching for more information because is some kind on mix in your post. In GenAI you always have a foundation model. The most widely used is ChatGPT. The foundation model contains all the knowledge. There are others. To have your own foundation model is quit impossible because the astronomic cost to maintain it. For this reason, you can see that models like ChatGPT are updated up to one specific date. With that said, the architecture to use it can vary. The most used is SaaS but you can use on-premisse. The las option is not usually to use ChatGPT, it is because on the foundation model you can put your own knowledge base. Here is the point where again SaaS or on-premisse have to be decided. So, my recommendation is search for architecture. For example inside the Accenture web site.
...
1 reply by Mohamad Ali Khatoun
Mar 17, 2024 3:31 PM
Mohamad Ali Khatoun
...
Thank you, Sergio, for your feedback.
I will firstly clarify my intent.
My query emanated mainly from the challenge of “data privacy”: How to feed Chat GPT, requesting solutions, while still keeping the company’s data private?
I was looking for solutions used by PMs to go around this challenge: Specifically, if anyone has adopted options 2 or 3 above.
Senior Projects Manager | Field & Marten AssociatesNew Westminster, British Columbia, Canada
Mohammad, we're still workign within option 1 for now and didn't explore the other two options!
...
1 reply by Mohamad Ali Khatoun
Mar 17, 2024 3:35 PM
Mohamad Ali Khatoun
...
Thank you Rami.
I hope a fellow PM will share his or her "exploration" of options 2 or 3.
Saving Changes...
Mike FrenetteManager, IT PMO| Halifax Water (retired)Halifax, Nova Scotia, Canada
ChatGPT for personal purposes and, for business purposes, Microsoft Copilot in the organization's data center or cloud tenant seems to be a model many organizations follow. I think we will eventually examine options that deal with SaaS and Large Language Models. Privacy and security are often issues that impede progress in these areas.
...
1 reply by Mohamad Ali Khatoun
Mar 17, 2024 3:42 PM
Mohamad Ali Khatoun
...
Thank you Mike.
I find Copilot a powerful addition from MS, navigating around its products. ChatGPT-like solutions cater for different needs, and that is what i am after.
Where have you benefited most from Copilot?
First of all sorry if I did not understand well your post. Please let me say my recommendation is searching for more information because is some kind on mix in your post. In GenAI you always have a foundation model. The most widely used is ChatGPT. The foundation model contains all the knowledge. There are others. To have your own foundation model is quit impossible because the astronomic cost to maintain it. For this reason, you can see that models like ChatGPT are updated up to one specific date. With that said, the architecture to use it can vary. The most used is SaaS but you can use on-premisse. The las option is not usually to use ChatGPT, it is because on the foundation model you can put your own knowledge base. Here is the point where again SaaS or on-premisse have to be decided. So, my recommendation is search for architecture. For example inside the Accenture web site.
Thank you, Sergio, for your feedback.
I will firstly clarify my intent.
My query emanated mainly from the challenge of “data privacy”: How to feed Chat GPT, requesting solutions, while still keeping the company’s data private?
I was looking for solutions used by PMs to go around this challenge: Specifically, if anyone has adopted options 2 or 3 above.
So, do you know of anyone who adopted AI under the SaaS or on-premise? or is it too early? How about the “sovereignty” of company data?
...
1 reply by Sergio Luis Conte
Mar 21, 2024 5:37 AM
Sergio Luis Conte
...
You are welcome. I know lot of companies. In fact, you will find lot of use cases outside there. The key point is cost and privacy. What is mostly used are hybrid models where the fine tunned component is inside the company that need to expose its internal knowledge to users.
ChatGPT for personal purposes and, for business purposes, Microsoft Copilot in the organization's data center or cloud tenant seems to be a model many organizations follow. I think we will eventually examine options that deal with SaaS and Large Language Models. Privacy and security are often issues that impede progress in these areas.
Thank you Mike.
I find Copilot a powerful addition from MS, navigating around its products. ChatGPT-like solutions cater for different needs, and that is what i am after.
Where have you benefited most from Copilot? Saving Changes...
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
Mar 17, 2024 3:31 PM
Replying to Mohamad Ali Khatoun
...
Thank you, Sergio, for your feedback.
I will firstly clarify my intent.
My query emanated mainly from the challenge of “data privacy”: How to feed Chat GPT, requesting solutions, while still keeping the company’s data private?
I was looking for solutions used by PMs to go around this challenge: Specifically, if anyone has adopted options 2 or 3 above.
So, do you know of anyone who adopted AI under the SaaS or on-premise? or is it too early? How about the “sovereignty” of company data?
You are welcome. I know lot of companies. In fact, you will find lot of use cases outside there. The key point is cost and privacy. What is mostly used are hybrid models where the fine tunned component is inside the company that need to expose its internal knowledge to users. Saving Changes...
Community Champion
