Voiceflow PartnersVPVoiceflow Partners
Powered by
BrandersnatcherB
Voiceflow Partners•2y ago•
4 replies
Brandersnatcher

Is it normal for each AI response to use around 1400 - 1800 tokens?

Just wondering if I should be using the knowledge base query api, or if using the KB response in the UI is the same thing.

I'm a bit nervous about launching my bot and destroying the quota when our users start trying it out.
Voiceflow Partners banner
Voiceflow PartnersJoin
Exclusive community for Voiceflow Solutions Providers & Content Creators.
13,076Members
Resources
Recent Announcements

Similar Threads

Was this page helpful?
Recent Announcements
jacklyn

# Let your agents search the web 🔎 Hey everyone! We just released the web search tool for the agent step! Your agents can now automatically search the web for information, letting your agent supplement the LLM's knowledge and the data in its knowledge base with live, up-to-date information. Plus.... * You can restrict searches to specific domains, so your agent only searches sites that you own * This is a tool, so you remain in control of when the agent searches the web * Results are automatically summarized in a way that your agent can automatically understand Under the hood, we're using OpenAI's web search API. Give it a try, and let us know what you think! https://docs.voiceflow.com/changelog/native-web-search-tool

jacklyn · 4mo ago

jacklyn

jacklynbiggin's Thread

jacklyn · 4mo ago

hurt-tomato

connor_maclean's Thread

hurt-tomato · 4mo ago

Similar Threads

Unresponsive "Response AI "
moderate-tomatoMmoderate-tomato / ❓┃ask-a-question
2y ago
Empty response each API request?
dangerous-fuchsiaDdangerous-fuchsia / ❓┃ask-a-question
3y ago
Hallucination in Response AI
dual-salmonDdual-salmon / ❓┃ask-a-question
2y ago
Memory and Prompt for Knowledge Base Response AI
opposite-copperOopposite-copper / ❓┃ask-a-question
2y ago