First off, I want to express my appreciation for the Genesys DEV team and the incredible work they've put into creating the Virtual Agent. It's a powerful tool with huge potential for transforming customer service.
While the idea behind Virtual Agents great and meant to reduce costs by handling calls that would otherwise go to human agents, the current pricing model seems counterintuitive, especially for high-volume scenarios.
This cost seems excessively high and defeats the purpose of using a Virtual Agent to reduce the cost of the operation.
Original Message:
Sent: 12-02-2024 18:04
From: Anton Vroon
Subject: Virtual Agent Value Proposition
Now that Virtual Agent has been out a bit longer, and we have a few new features out. Has anyone's opinions changed on VA.
This week we are getting Article Summarization and Improved slot filling using LLMs (though would love to see this in the base product, especially for voice bots)
How are people feeling about the value of VA now, is it worth the cost that is being asked for?
For me, personally, still seems really high cost for what we get to me at least. Maybe the future state might be worth the cost, but also feels like if I just want to do some phrase generation for Intents, that is way to much to ask, and is all in or nothing.
------------------------------
Anton Vroon
Original Message:
Sent: 11-04-2024 16:16
From: Anton Vroon
Subject: Virtual Agent Value Proposition
I'm trying to wrap my head around the costing for Virtual Agent
2 sessions per token, where each interaction that goes through the virtual agent is 1 or more session. Lets be generous and say 1 session per call. 1$ USD for 1 Token
1,000,000 interactions a year that you want to provide virtual agent capability for, that is $500,000, which seems really steep compared to the other AI features in Genesys. - And yes I understand it is free for a limited time.
Especially since this is a new product, and what we have right now is only, ok.
So what do we get?
Generate intents and utterances. This should be free, and not tied to Virtual Agent. I can use any LLM like ChatGPT, ChatSonic or Claude for free to generate phrases for multiple intents at the same time, and I just need to copy and paste it over, so a minor addition to remove the copy paste.
Other LLMs like Claude do a much better job of it, able to build upon the phrases generated, I was genuinely impressed with it's ability to localize the phrases to NZ using colloquialisms and even Te Reo in places. Tried the same with Genesys utterances, and it wasn't even close.
Yes Claude is a top tier LLM, but the point being it is free and more capable, so paying for this at this price for Genesys seems a bit much.
Now yes Virtual Agent has other capabilities, but so does Claude which can also generate and modify and error check call flows as a YAML file that can be uploaded with Archy, for free, I've only done really basic stuff with that but it will get better.
What else?
Auto wrap up codes - we could do with Data Actions already so minor improvement
So what we are really paying for is:
Knowledge highlight
Knowledge generation based on Knowledge articles
Conversation Summary - While with Virtual Agent
I understand more is coming for Virtual agent, But I'm just wondering is that worth it at that price tag?
Hoping that maybe someone can convince me there is a good ROI here that I'm not seeing.
I am excited about the use of AI, but I'm just not loving this one yet,
#ConversationalAI(Bots,AgentAssist,etc.)
------------------------------
Anton Vroon
------------------------------