Description
Summary We've built and validated the front end of a paid psychometric assessment — a 50-question Highly Sensitive Person (HSP) test running on Typeform, with scoring already calculating correctly and landing in Google Sheets. What's missing is the "brain" behind it: an automated system that takes those raw scores, makes sense of them the way a psychologist would, and sends the client a beautifully written, personalized report. Since real customers are paying for this, the bar is high: the system needs to work the same way every single time — no random variation, no broken formatting, no visible code or JSON leaking into what the customer reads. The Build A 3-step AI agent chain: Step 1 – Data Checker: catches missing values or inconsistent scores before anything moves forward Step 2 – Interpreter: reads the scores against our clinical framework and drafts the psychological interpretation Step 3 – Writer: turns that draft into a polished, human-readable report — clean text only, no artifacts Tools We're Expecting to Use Agent orchestration: CrewAI, LangGraph, or LangChain Claude 3.5 Sonnet for the empathetic clinical writing, GPT-4o (Structured Outputs) for reliable parsing n8n as the automation backbone (Make.com is fine too if you have strong reasons to prefer it) Temperature locked at 0.0, few-shot prompting, tightly controlled system prompts — consistency is more important than creativity here What "Done" Looks Like Stripe payment + Typeform submission automatically kicks off the whole pipeline Final email report arrives via Outlook, fully formatted, with zero technical leakage You leave us with simple documentation so we can tweak prompts or rotate API keys without needing you every time Who Should Apply If you've actually shipped multi-agent AI systems before (not just experimented with them), and you're comfortable with n8n/Make plus structured LLM output — we want to hear from you. Bonus if you've worked on anything in the assessment, coaching, or mental health tech space. To Apply Share 1-2 examples of AI agent pipelines or structured-output systems you've built and shipped to production.