AI Automation Developer for Email, Chat & Phone Support - AI Agents

We are a telemedicine brand looking to automate customer support across email, messaging portal, and phone. The goal is to resolve 80%+ of inbound inquiries without a human rep — cutting ticket volume dramatically so our team can focus on complex cases.

A note on where AI is today: The landscape has changed dramatically. AI agents can now live on a computer, browse platforms, take actions, and make decisions autonomously — things that would have required months of custom engineering two years ago can now be set up or built in days. We're looking for someone who is deeply plugged into these advancements and knows how to leverage them to keep costs low and deployment fast.

The Challenge

This isn't a simple FAQ bot. Most of our support tickets require looking up live patient data before responding:

"Where is my order / what's my tracking number?"

"What dose am I on? What did my doctor prescribe?"

"When does my subscription renew? Can I pause it?"

"Has my prescription been sent to the pharmacy?"

Right now a human rep logs into our telemedicine platform, finds the patient, pulls their order/prescription/subscription details, and then responds. We need AI agents to do this automatically — identify the patient, look up their info in real time, and respond accurately.

Important: Our telemedicine platform has no API access. Any data lookup will require browser/computer-use automation.

What We Need Built

1. Email (Zendesk)

Classify intent → identify patient → look up their data via browser automation → send an accurate, personalized response automatically. Escalate to a human for clinical or complex issues. This can be built natively within Zendesk AI or as an external integration — we're open to whatever approach works best.

2. Messaging portal

Same data-lookup logic, but in a messaging portal interface.

3. Inbound Phone (AI Voice Agent)

Natural-sounding voice agent that answers calls, verifies the patient, looks up their account in real time, and handles common inquiries without transferring to a human. Hands off to a human if it cannot answer the query or gets stuck. If human phone agent not available, it leaves a message for our human reps to follow up on.

The AI should learn from our historical emails and messages to understand how we respond to different cases and replicate that tone and logic.

Who We're Looking For

Experience building AI agents with LLMs (GPT-4, Claude, Gemini, or similar)

Browser/computer-use automation (Playwright, Puppeteer, or similar)

Zendesk integration and/or Zendesk AI setup

Voice AI platforms (Bland.ai, Vapi, Retell AI, or similar)

Agentic frameworks (LangChain, CrewAI, or similar)

Experience with sensitive data / healthcare-adjacent environments a plus

Open to a freelancer or small team. Also open to a phased rollout (email first, then chat, then phone).