Hi Retell team,
We are Proto Health, and we are currently building an AI calling integration for a pilot the week of May 4.
To meet our pilot requirements, patients must see the clinic’s own number as the caller ID. Additionally, the clinic’s PBX must be able to transfer inbound calls to a Retell agent while preserving patient context in SIP headers. We have several technical questions regarding outbound BYOC and inbound SIP that are blocking our integration decisions before the April 25 trunk provisioning cutoff:
Outbound BYOC Caller ID
1. Does Retell perform number-ownership verification (LOA, CNAM lookup) for BYOC SIP-trunk imports, or is trust based on the trunk accepting the call?
2. On outbound calls via an imported trunk, does Retell set the P-Asserted-Identity / From headers, or pass through the trunk’s settings?
3. Can we dial out from a US-imported number with a different display-only override or CNAM?
4. What is the current production turnaround time for Verified/Branded Caller ID business-profile approval in the US?
5. For the “Dial to SIP URI” path (registerPhoneCall), does the caller ID reflect our upstream trunk’s setting (e.g., Twilio), or does Retell override it?
6. Are there specific concurrency or rate limits for imported SIP trunks compared to Retell-provisioned numbers?
Inbound SIP
7. Does Retell accept inbound SIP INVITEs from a third-party PBX? If so, what endpoint and authentication methods (IP allowlist, digest auth, mTLS) are supported?
8. If a third-party PBX transfers a call with custom SIP headers (e.g., X-Patient-ID), are these exposed to the agent runtime as dynamic variables for the LLM?
9. Can a Retell agent hand off a call mid-conversation to an arbitrary SIP URI? Does a built-in transfer tool exist, and does the handback preserve the original caller’s PSTN identity?
Since these features are blocking, we would appreciate your feedback soon so we can finalize our provisioning.