We are researching how configurable answer generation for Knowledge Fabric could work in the product, and we would love input from the community.
We are especially interested in the real-world use cases where you would want to tailor generated answers, such as:
-
Length - when do you need short vs. detailed responses?
-
Tone - should answers be formal, conversational, empathetic, instructional, or something else?
-
Style/format - do you need bullets, step-by-step guidance, email-ready responses, executive summaries, or other structured outputs?
-
Audience - are the answers intended for admins, agents, supervisors, end customers, or internal teams?
-
Channel/context - would this be most valuable in Agent Copilot, Virtual Agent, customer self-service, internal knowledge search, or elsewhere?
-
Guardrails - what controls would you want to make sure responses stay accurate, compliant, and on-brand?
We are also interested in how you would expect to provide and manage these configurations in the product. For example:
-
Should this be set through simple presets, editable instructions, dropdown options, or more advanced policies?
-
Would you want configuration at the global level, within AVAs, by knowledge configuration, in AI Studio, etc.?
-
Would previewing and testing responses before publishing be important?
A few questions that would help us as we evaluate this idea:
-
What specific scenarios or workflows would benefit most from configurable answers?
-
What kinds of answer controls would be most important for your team?
-
How would you expect to set up and maintain those controls in the product?
-
Would you want these settings applied globally, by knowledge source, by assistant/bot, or per use case?
-
Are there examples of answers today that are technically correct but not in the right format, tone, or level of detail?
-
Are there any industry, compliance, branding, or localization requirements we should consider?
We also have an Aha! idea on it too: https://genesyscloud.ideas.aha.io/ideas/SSAKF-I-323
#AIConfiguration
#AICopilot(Agent,Supervisor,Admin)
#ConversationalAI(Bots,VirtualAgent,etc.)
------------------------------
Prateek Garg
Staff Product Manager, Knowledge Fabric
------------------------------