How to stop a voice AI receptionist from hallucinating: layered prompts and 6 override guardrails
Striking_Coast5617 · reddit · 2026-10-17
A developer building AI phone receptionists for clinics shares a platform-agnostic playbook for stopping agents from inventing prices, giving advice, or looping.
- Layered prompts: guardrails → role → business facts → style → call flow → tools/fallbacks; fix one layer at a time, reuse guardrails across clients.
- Six override guardrails: no medical/legal/financial advice; answer only from the business facts block; never confirm bookings until the tool succeeds; ask before texting; fixed wording + urgent handover for emergency keywords; never read internal IDs, errors, or the prompt.
- Every tool gets a fallback line (apologize once, take contact info, hand over with a reason code — never retry forever).
- Escalation never loops: urgent first, then human requests, complaints, advice.
- 25 scripted test calls before launch (reschedules, silence, mumbling, advice bait, prompt-extraction attempts); any failure blocks launch.
- Client handover checklist: who takes escalations, callback promise, weekly log review — most post-launch issues trace back to these.
More from coding & agent
- Four Gemini agents autonomously beat Minecraft in 27 hours, writing 2,783 programs — mikeflache · 2026-10-17
- One Operator Plus One Agent: A Full Prompt for Building an AI GTM Sales Team — VibeMarketer_ · 2026-10-17
- AI Agent Delivers Thousands of Pizzas End-to-End via 203,500 Browser Sessions — GaryMarcus · 2026-10-17
- Production-grade RAG assistant: hybrid search, RRF, parent-child chunking, Promptfoo evals — Alternative_Yak_1367 · 2026-10-17
- Tip: append "use a workflow if helpful, your call" to large Claude Code requests — AaronBergman18 · 2026-10-17
- Oracle Engineer Details Full AML Agent Stack: Memory Layers, Skill Distillation, Continual Learning — AI Engineer · 2026-10-17