Using Additional Instructions to Handle Buying Intent in a Chatbot

While working on Fabio AI Chatbot, we were testing ways to customize responses without touching the main prompt architecture.

Originally, the Additional Instructions field was simply meant to adjust chatbot behavior.

But during testing, we noticed something interesting:

It can also be used to define reactions when the chatbot detects buying intent patterns.

Not to change answers.
Not to make the chatbot pushy.
But to influence what next action becomes visible when a visitor already shows intent.

Examples

Visitor asks about implementation
→ indicate appointment booking

Visitor compares plans
→ indicate pricing or software download

Visitor asks how to start
→ indicate course registration

What we liked about this approach:

• No custom model training
• No extra API calls
• No intent classification service
• Uses an existing configuration layer
• Website owners remain in control of behavior

To make experimentation easier, we prepared 20 copy-paste buying intent templates.

The templates are intentionally generic so they can work across multiple use cases:

• SaaS
• Agencies
• Online courses
• Service websites
• Software downloads
• Lead generation

If you are building AI features, I’d be curious:

Would you implement intent handling directly in the system prompt, through configuration layers, or through a separate intent engine?

Templates:
https://fabio-plugins.com/buying-intent/