Make your Assistant agentic
Let's make our AI assistant agentic. It can decide whether to give a quick answer or dive deeper when it's asked information for a specific customer.
So far, your AI assistant can either answer a general sales policy question or stay silent. In this lesson, you’ll give it more power by teaching it to recognize when a user is asking about a specific customer and to pull in extra context from other tools like your CRM.
Here’s what you’ll set up:
- New Router route: Add a third option so the assistant can recognize sales policy questions that also mention a customer email.
- Extract data: Use the Extract Data node to grab details (like an email address) directly from the message.
- Subflows for external tools: Send that email into a subflow that fetches details from your CRM (e.g., HubSpot) and compares them to your sales policies.
- Return the analysis: The AI will combine CRM info with policy rules and return a clear decision back in Slack (or wherever the assistant is deployed).
By the end of this lesson, your AI assistant will have three tools at its disposal: do nothing, answer a general question, or run a deeper analysis for a specific customer. From here, you can keep adding tools and routes, expanding what your assistant can handle until it becomes a true AI teammate for your team.
