Run your conversations through the model you chose in AWS Bedrock and act on what it finds.
Get more done with Tidio and Levity integration
Use a Levity model to classify Tidio conversations and route them without reading each one.
No credit card required
Contact sales
Integration availability depends on your Tidio plan.
Compare plans
Category
AI & LLMs
Integration availability depends on your Tidio plan.
Compare plans
Use cases
What you can do with Tidio and Levity
Triage the queue automatically
A Levity label on the first message sorts refund, billing and bug conversations into the right queue.
Route by intent, not keywords
Keyword rules break on phrasing. A trained model handles the ways people actually write.
Tag conversations for reporting
Consistent labels make the monthly picture of support something you can compare.
Classify what customers send
A photo or document sent in chat can be classified before an agent opens it.
Overview
Levity lets you train a classifier on your own examples: this is a refund request, this is a bug, this one is angry. Support conversations are exactly the kind of text that benefits. Sending the first message of a Tidio chat to Levity and using the label to route or tag it means the queue sorts itself, with a model trained on your categories rather than somebody else's. Low confidence predictions can go to a human, which is the part most setups forget.
Key features
- Your categories, not generic ones A Levity model is trained on your own conversation examples.
- Routing from the first message The label decides the queue before anyone has read it.
- Confidence you can act on Uncertain predictions go to a person instead of a guess.
- Images and documents too Levity classifies an uploaded receipt or photo, not just text.
Levity
AI & LLMs
Tags
Custom Integration
Workflow Automation
How it works
Connect Levity in minutes
1
Connect Levity and Tidio
Zapier, Integrately or Make can send messages for classification, or the Tidio API can post the label back.
2
Train on real conversations
Export a few hundred labeled chats so the Levity model learns your categories.
3
Decide the confidence rule
Above your threshold the label routes. Below it, a person reads the message.
Discover more integrations
All Integrations
Read the mood of a conversation, sort it by topic, and keep personal data out of your logs.
Score the person in the chat with your own model and treat them accordingly.
Keep website chat and student outreach telling the same story, in both directions.
Use your custom GPT in live chat to handle more conversations your way.
Route hard questions to Claude and send the answer back into the Tidio conversation.
Looking for customer service automation that fits your business?
We build custom API integrations to streamline your most important workflows.
Frequently asked questions
Common questions about the Levity integration
How many examples does a Levity model need?
Enough per category to cover the ways people phrase it, which in practice is a few dozen at minimum and a few hundred to be comfortable. Start with your three or four most common categories rather than a taxonomy of twenty, because a model spread thin gets everything slightly wrong.
What happens when the model is unsure?
Set a confidence threshold and send anything below it to a human queue. That single rule is the difference between a classifier that saves time and one that quietly misroutes the awkward cases. Levity returns the score, so the rule is yours to set.
Is the Levity integration available on my plan?
Availability depends on the Tidio plan you are on. Compare plans to see which plan covers the Levity integration.