AI Implementation Analyst
About the company
Our client is a startup building an AI agent that handles patient communication for healthcare clinics: scheduling, pricing, doctor availability and day-to-day questions.
About the role
The AI agent is only as good as what it knows, and this role owns that knowledge.
Every clinic works differently: different doctors, prices, schedules, rules, exceptions, and things everyone on staff knows but nobody has ever written down. The person in this role will get all of that information out of the clinic, structure it so the agent can use it, test it against real cases, and refuse to launch until it's right. After launch, they will read real patient conversations, find what's wrong before the clinic does, and fix it.
This is a craft role, not an engineering one. It requires unusual precision and the patience to get the last details right when everyone else is ready to call it done.
What this person will own
Build speed: how fast a clinic goes from information gathered to an agent that passes its tests
Launch quality: nothing goes live wrong
The agent's accuracy in production
What this person will do
Build it right
Turn how a clinic really operates into an Agent Knowledge Base the AI agent can act on. This is the core craft of the role
Get information out of people who don't know they have it, like the exception the receptionist handles from memory or the rule that only applies on Saturdays
Run acceptance tests, and hold a launch when it doesn't pass, even when the date is uncomfortable
Keep it accurate
Read real patient conversations and find the failures nobody reported
Classify what went wrong, fix what falls within their scope, escalate what doesn't, and make sure the same issue never comes back
Bring Product and Tech evidence, not anecdotes
Make the next one cheaper
Turn everything repeatable into automation, defaults and documentation that a CSM can run independently
Build the standard, so the next clinic doesn't start from a blank page
The company measures how many of this person's hours each new clinic costs, and that number is expected to fall over time
Profile
1–3 years of experience in implementation, onboarding, operations, QA, technical support, or any role where they personally owned whether something was correct
Meticulous in a way that mildly annoys people: they notice the contradicting sentence, the wrong price, the edge case nobody asked about
Writes with precision. Ambiguity bothers them, and they'd rather rewrite a sentence four times than leave it open to interpretation
Learns new tools fast and enjoys configuring things properly. Comfortable with data and able to work out what happened without someone explaining it
Would rather hold a launch than ship something they know is wrong, and says so out loud
Solves for the second clinic, not just the one in front of them
Follows a process without being chased, and speaks up when the process is wrong instead of working around it
Experience with AI agents, chatbots, or healthcare operations is a plus, not a requirement