Why customer would potentialy use other models

[User687957]

Dear all,

During my exploration of Now assist and Agent AI, I have seen that users can potentially use theire own LLM model instead of the ones embeded into serviceNow.

I have a wierd question about this .

Why customer would need to use their own model as most of the work in Service now has been done already ?

1. might be link to customer company rule on validated internal models
2. Could it be for having full control of their own cost
3. Could it be for specific operation

Thanks for sharing on this

regards

[soumyadeep10]

Hi @User687957 ,

1. ServiceNow's LLM is totally based on ServiceNow's features but when there is a need for some specialized tasks be it content generation, sentiment analysis, data analysis, etc., we might need our own LLM.

2. Organizations may need a specific model that Now LLM does not provide. This might be GPT-4 for reasoning or a smaller model for speed.

3. Custom LLM models are more versatile and can handle multiple use cases beyond the ServiceNow platform.

4.  If your use case is tightly bound to ServiceNow’s ITSM, HR, or CSM workflows, Now LLM is the logical choice. If the goal extends beyond ServiceNow’s ecosystem, consider a custom LLM or a hybrid approach. 

*************************************************************************************************************
If my response proves useful, please indicate its helpfulness by selecting " Accept as Solution" and " Helpful." This action benefits both the community and me.

[yashkamde]

Hello @User687957 ,

There reason depends on Unique Business Context and Proprietary Data because servicenow’s pre-built models are trained on general IT or HR data. However, an organization's proprietary, niche data is what creates a competitive advantage.

Also High-Precision Requirements & Trust many organizations face a decline in trust surrounding AI generated outputs, prompting them to take control over model training to ensure quality and reliability.

In short : ServiceNow provides the framework, but customers use their own models to inject their domain expertise and proprietary data to move from general automation to specialized, high-value AI applications.
Basically it's about governance, flexibility, and tailoring AI to their business context.

If my response helped mark as helpful and accept the solution.

[rpriyadarshy · Accepted solution]

@User687957 

Here is Pointed Answer on Your Queries.

1. might be link to customer company rule on validated internal models--> Many Customer prefer Own model to have better control over and Company Specific AI Guardrails. With Now LLM its restricted to use settings which are just exposed in form of Properties.
2. Could it be for having full control of their own cost--> Cost I dont think a big angle but yes. Now a days many customers has Team of LLMs n SLMs and Based on performance and Cost it can be Used.
3. Could it be for specific operation--> Ever model has its own performance based on Use case. As some models are good in Image Processing , Some are good in Summarization, Some are good in Voice to Text etc etc so Based on Use case we can use different models . So Yes its based on Use case Also.

Details are here on BYOLLM or BYOK

https://www.servicenow.com/community/now-assist-articles/using-external-llms-with-now-assist-includi...

Regards

RP