Our AI has handled millions of Instagram DMs for high-ticket coaches, and it's been attributed to over $20M in sales so far.
The goal now is to make it the best sales brain in the world, and I need someone to make sure it only ever gets better.
Quick context: SettyAI runs DM sales for high-ticket coaches. It's in the conversations setters usually need years to get good at. "I need to think about it." The guy asking the price two messages in. The silence after you send the offer. We're teaching it to deal with what's actually behind the objection, before the buyer even says it.
The way we see it is that every time we fix something, every bot should get smarter.
So one client's win should become every client's win and that only works if changes are safe to push, which is what this role is about...
The role: Full Time Prompt Engineer (LLM testing + feedback loop)
Why it's a genuinely hard problem:
- If something breaks, our clients lose money
- Our prompting systems are up to 180k characters, running live across dozens of businesses
- Most eval tools are built for simple chatbots, not long sales conversations with buyers who say whatever they want
- Anything you improve goes out to every client at once so the leverage is big
What you will own:
- The feedback loop. When something breaks for a client, or we find a better way to do something, you decide if it goes into the base and you push it out to everyone
- The testing that makes that safe. Nothing reaches a client without being tested automatically first, and nothing that worked before breaks
- Simulated prospects that actually behave like real buyers. Messy, unpredictable, different every run, not the same script on repeat
- Working out pass/fail when every conversation is different
- The base prompt itself. Figuring out why the model does what it does and fixing the actual cause
To be clear, this is prompt work, not infrastructure or general software engineering.
You're probably a fit if:
- You've built automated testing or evals for full LLM conversations, not just single replies
- You've done prompt engineering in production, where getting it wrong cost money
- You'd rather own an outcome than get handed tickets
- Your English is fluent, written and spoken
Bonus if you've worked on conversational agents, voice agents, AI sales tools, or you know the info space.
What you get:
- Full ownership. How you do it is up to you, we care about results
- You work directly with me and my co-founder. No layers, decisions happen fast
- Room to grow into running the whole function as the team grows
- Up to $80K/year, plus equity for the right person
- Remote, full-time, long-term
- We're an official Meta partner, growing fast, and we want to be the default way online businesses turn DMs into customers
