Define the intents (e.g. sales, support, feedback, general questions).
Draw a flow per intent.
2
Step 2: Configure the trigger and input
In the bot builder, set the trigger (incoming message, keyword, etc.).
Add an AI / logic block to analyze the intent.
3
Step 3: Configure intent detection
Use intent detection (often NLP) to classify messages.
Add example phrases per intent to improve accuracy.
4
Step 4: Route according to intent
Route to the correct block or agent (sales to sales, support to support, etc.).
Add a fallback for ambiguous cases, with a clarification question.
5
Step 5: Connect trained agents
For document-based responses, connect the relevant AI agent.
Check that it is trained on the right files.
Configure response mapping and a fallback.
6
Step 6: Add escalation to a human
For unresolved cases (or complaints), escalate to a human agent.
Add a clear transition message.
7
Step 7: Test and optimize
Test with varied messages.
Adjust intents, add examples, optimize fallback and escalation.
If you provide a transcript or detailed steps from the video, we can produce a more precise guide. For now, these steps describe a standard intent detection and routing flow.
