Automation · Setup in 10 min· Setup Pass

Set up Decagon to answer client FAQs in 10 minutes

Configure an AI agent to handle common client questions about your practice, protocols, and scheduling.

Repetitive client questions about logistics, standard protocols, or service details can consume significant front-desk time. This guide configures a Decagon agent to provide instant, accurate answers, letting your team focus on higher-value client interactions.

Before you start

• A Decagon account
• A document with your top 20 client FAQs
• Access to your clinic's knowledge base or website

The steps

1. 01

   Connect your knowledge base

   First, point Decagon to your existing documents. Upload your FAQ document, service descriptions, and any standard pre-procedure instructions directly. If you have a public website with this information, you can also provide the URL for Decagon to crawl.

2. 02

   Create the "Front Desk" agent

   In your Decagon workspace, create a new agent and name it "Front Desk." This agent will be the primary contact for incoming client queries. Define its personality as "Polite, concise, and helpful," and instruct it to only answer questions based on the provided knowledge base.

3. 03

   Define query-handling rules

   Establish rules for when the agent should escalate a query to a human. A good starting point is to escalate any question containing words like "emergency," "symptoms," "allergic," or any mention of a specific medical condition not covered in your standard protocols. This ensures clinical nuance is handled by a practitioner.

4. 04

   Configure a sample client protocol query

   Test the agent by asking a question a client would, such as, "What supplements do I need to stop taking before my blood draw?" The agent should pull the answer from your knowledge base. This is the Ledger layer of your client support, providing a consistent record of information.

5. 05

   Test the escalation path

   Type a query that should trigger escalation, like "I'm feeling unwell after my treatment." Verify that the agent responds with a message like, "I cannot answer medical questions. A member of our team will contact you shortly," and that the query is correctly routed to your front-desk team.