Making a phone agent’s behavior configurable and demonstrable.

A business needs to understand how an AI phone agent will behave before putting it in front of callers. The opening line, voice, instructions, and limits are all part of that behavior. A general conversation demo does not show how those settings apply to a business task.

TalkPatch is a self-directed demonstration of that configuration experience. A dental receptionist scenario gives the settings a concrete context: handling appointment requests, recognizing urgent calls, and knowing when a person should take over.

I built the product experience for defining agent behavior, opening lines, voice settings, and business-specific guardrails across inbound and outbound call workflows. The project includes voice AI configuration, ElevenLabs, and webhook integration work.

I used the product to demonstrate the dental receptionist flow. The scenario shows how a configured agent can move through different caller needs rather than treating every call as the same conversation.

Appointment requests, urgent-call triage, and human handoff are different paths with different responsibilities. Making those paths visible helps a buyer evaluate what the agent is intended to do and where staff remain involved.

The demo is an inspectable product example. A production deployment would need the practice’s own scheduling integration, escalation process, and operational validation before handling real patient calls.

TalkPatch demonstrates agent configuration and a business-specific receptionist workflow. The public demo and portfolio walkthrough are linked below.

This is independent demonstration work. It does not represent a commissioned dental deployment or a measured reduction in missed calls, staffing costs, or booking losses.