Building Outbound Voice Agents at Scale - DIDs, Spam Filtering & Multi-Client Architecture
Hey everyone, I'm running an AI voice agent for insurance outbound. The agent works well, but my telephony layer failed. Using one number got spam-tagged fast. Buying fresh Twilio numbers didn't help, they were instantly marked spam/not-connect. I'm not looking for "use more numbers" advice. I want the playbook for doing outbound at scale without burning DIDs. Specifically: 1. Trust/Reputation - How do you build/maintain trust score + avoid spam labeling across carriers? - Is branded caller ID (Hiya/First Orion/TNS) mandatory now? 2. Number Pool Management - What's a realistic DID-to-calls/day ratio? - Rotation rules that actually work? - Local presence vs toll-free: what's performing better? 3. Setup Per Client I want to offer this to multiple clients. I assume each brand needs: - its own TCR/A2P registration - its own DID pool - its own attestation level - subaccount architecture under my master account Is this the right structure, or should every client have their own full provider account? Any architecture diagrams or benchmarks for answer rate trends would be huge help. Thanks.