AI Automation Specialist, Customer Experience

About UsHyprwork is a fast-growing operator of direct-to-consumer brands in health and wellness. Our flagship brand, Rejuvacare, has scaled 12x in a single year to nearly 150 people across 15+ countries. We ship primarily to the United States and serve hundreds of thousands of customers who trust us with their health. Every customer interaction is a brand moment, and we are building the AI infrastructure to make every one of those moments better, faster, and more consistent.

The RoleThis is not a customer service role with AI sprinkled on top. This is a technical role for someone who builds, ships, and iterates on AI-powered systems, and whose domain happens to be customer experience. Hyprwork's CX operation serves a large and growing customer base across email, phone, and chat channels. The current setup runs on Richpanel and Aircall, supported by a team of human agents. The problem is efficiency: resolution times are too long, containment rates are too low, and the team cannot scale linearly with customer growth. The solution is an AI-first CX layer that handles routine interactions autonomously, routes complex cases intelligently, and gives human agents the tools to resolve issues faster when they do step in.

You will design, build, and maintain the AI automation layer that powers this transformation. That means writing and tuning the prompts that drive AI agents, building evaluation frameworks to catch failures before customers do, managing integrations between Claude, Richpanel, and Aircall, and translating CX performance data into system improvements. You work alongside the AI Technical Manager and partner closely with the CX operations team, but your output is technical: prompts, workflows, integrations, evals, and automation logic.

This role exists because the CX team needs to get dramatically more efficient, and the path to that efficiency runs through AI and automation, not through hiring more agents.

What You Will Own AI Agent Design and Prompt Engineering Writing, testing, and iterating on the prompts that power AI-driven customer support across email, chat, and voice channels. Building and maintaining the knowledge base architecture that feeds AI agents: product information, policy documents, troubleshooting guides, and tone guidelines. Designing conversation flows that resolve customer issues autonomously where possible and route to human agents intelligently where necessary. Tuning AI responses for accuracy, tone, compliance, and resolution quality across different interaction types.

Evaluation, Testing, and Quality Systems Building evaluation frameworks and test suites that measure AI agent performance against defined criteria: accuracy, tone, resolution rate, and compliance. Running structured evals to catch regressions, hallucinations, and edge cases before they reach customers. Monitoring containment rate, CSAT, average handle time, and sentiment metrics to identify where the AI layer is working and where it needs improvement. Creating feedback loops between AI performance data and prompt iteration cycles.

CX Tech Stack Integration Managing and optimizing the Richpanel and Aircall configurations: workflows, routing rules, automation triggers, and integration points. Working with Claude (via API, Projects, or Claude Code) to build and maintain the AI support infrastructure. Partnering with the AI Technical Manager and engineering on integrations that connect the AI layer to CX platforms, internal tools, and data pipelines. Scripting and automating data extraction, reporting, and analysis workflows using Python or equivalent.

CX Intelligence and Continuous Improvement Analyzing call transcripts, chat logs, and ticket data to identify patterns: recurring customer issues, common failure points in AI responses, and opportunities for new automation. Translating CX insights into actionable system improvements: new prompt strategies, updated knowledge base entries, revised routing logic, or workflow changes. Documenting system architecture, prompt strategies, and operational playbooks so the AI layer is maintainable and improvable by the team, not just by you.