Hello Folks, We’re building a voice-based appointment booking agent for GP clinics in Europe using Vapi’s GPT-4o cluster through an n8n workflow. We’ve hit a critical blocker — the agent hallucinates or misinterprets time slot data when asked about multiple doctors in the same call. It often mixes up slots or gives completely wrong info, even though our backend API and workflow are returning the correct data. We’ve spent over 20 hours debugging, but the issue seems to lie with how GPT-4o is handling context across multiple queries in one call. We already sent a detailed email requesting urgent help and would really appreciate a Zoom call to troubleshoot this with your team directly. We’re ready to share our full setup and reproduce the issue live. This is urgent for us — we're close to going live with clinics, and reliability is key in a healthcare setting. Please help us sort this out 🙏 Thanks!