Field service techs, drivers, in-person sellers, anyone whose Tuesday is in a vehicle or on a customer floor: none of them are well-served by a chat window or a mobile app. A phone call is the only AI surface that’s safe and natural in those contexts. Henoyo provisions a real PSTN number through your Twilio or Amazon Connect account and runs the same governed Prompts your office team uses, hands-free, in under two seconds.
Phase 2 · Twilio first, Amazon Connect second.
It’s 8:40am. Diego is driving to a service call at the Acme Northbrook site in Pleasant Hill. Twenty minutes out. Hands on the wheel, eyes on the road. He hits the Henoyo number on his car phone.
Diego (over the phone, hands-free):“Pre-call brief on Acme Northbrook, the Pleasant Hill site.”
Thirty seconds later, on the speaker:
Henoyo (spoken):Pleasant Hill site, account is Acme Northbrook. Last service was an HVAC visit on March 12. Marcus replaced the chiller intake valve. The customer flagged a follow-up needed on that line two weeks ago. Two open tickets relate to it. The site contact is Pat Reyes; Pat prefers a five-minute heads-up text before arrival. The site also has a closing-time sensitivity. The warehouse shift change is at 4pm, so wrap before that or expect a longer wait at the gate.
Diego pulls into the lot at 9:00am. Pat meets him at the dock. Three hours later he’s back in the truck, merging onto the freeway. He hits the Henoyo number again.
Diego (over the phone):“Log the visit. I replaced the chiller intake. Closing both open tickets. Recommend a follow-up check in 30 days.”
Henoyo (spoken):Logging chiller intake replacement on Acme Northbrook · Pleasant Hill, closing tickets 8842 and 8859, scheduling a follow-up service for May 28. Confirm to write?
Diego says yes. The work order, the ticket closures, the follow-up: all in the system before he merges back onto the freeway. The audit trail captures Diego as the writing user, the field name, the before-value, and the after-value.
Three things, mostly.
Pre-call briefs, customer history, prior service notes, contact preferences: spoken back in thirty seconds, while the tech is driving. The 8am decision about which order to do the visits in stops being made on partial information.
“Log the visit” at the end becomes a thirty-second voice prompt instead of a forty-minute end-of-day chore. Skills that write to the CRM require explicit verbal confirmation. The audit trail captures who, what, and when.
No new device. No new app. No earbuds required. Bluetooth in the truck, AirPods on the walk, a regular call from the parking lot. Sub-2-second responses for short Prompts; 3–5 for fan-outs across multiple systems.
If you’ve looked at the major enterprise AI vendors for a field-team channel, you’ve found chat windows and mobile apps. Salesforce ships a mobile app. Microsoft ships Copilot in Outlook. ServiceNow ships a phone-friendly portal. None of them ship a meaningful AI-over-phone-call experience for the technician driving between job sites.
The reason is architecture, not effort. Voice-as-an-afterthought (bolting a TTS layer over a chat interface) produces the IVR experiences your customers complain about. A useful phone channel needs voice as the primary surface from the bottom up: ASR optimized for your domain vocabulary, Prompt outputs designed to be spoken, Skills that capture verbal confirmation, and latency budgets that respect a person who’s actually driving.
Henoyo was built voice-first. The phone channel isn’t a port. It’s the same architecture the in-browser voice channel runs on, plus the PSTN round-trip. Your field team is the team that needs AI most and gets it least. Telephony is how that changes.
Three example invocations: caller speaks the prompt, Henoyo speaks the response, write-back requires explicit verbal confirmation. Audit trail captures the call.
Customer history for the next visit, 2400 Maple.
Spoken readback: last service date, open work order, equipment under service, the prior visit note, the customer’s preferred contact name. Under thirty seconds.
Read logged in your Henoyo audit trail with the calling user, transcript, and timestamp.
Log the visit. Replaced the chiller intake. Closing tickets 8842 and 8859.
Henoyo confirms the write back verbally: “Logging chiller intake replacement, closing tickets 8842 and 8859. Confirm to write?” Then waits for an explicit yes before executing.
Write logged with caller, before/after field values, and the verbal confirmation transcript.
Order status check on PO 7715.
Spoken readback: current fulfillment status, latest committed ship date, any upstream delays, and the carrier reference number for follow-up.
Read logged in your Henoyo audit trail; transcript captured to the immutable log.
Same Prompts, every surface.
Three answers, in case you’re forwarding this page to your IT or telecom lead.
In your AWS or Azure account. The phone number is provisioned in your Twilio account or Amazon Connect instance, not on Henoyo. Calls terminate into the Henoyo container running in your cloud, which handles ASR, Prompt execution, and TTS. Tokens to your CRM stay in your secrets store. Call recordings, if you require them, stay with your telephony provider under your existing retention policy.
The Twilio webhook (or Connect contact flow) is configured by your team and points at the Henoyo container; turning it off is a console change in your telephony account. The container itself is also revocable independently from your cloud console. The phone number stays yours either way. Henoyo doesn’t resell numbers.
Recording is configurable per Prompt. If your compliance policy requires call recording (financial services, healthcare), recordings stay with your telephony provider under your existing retention policy. Henoyo never sees the audio file. Transcripts of every call are written to the same immutable audit log as every other surface, in your cloud, with six-year retention by default. Your compliance team gets one log to query across voice, telephony, Slack, Teams, and Copilot.
Twilio for Phase 2 (faster to ship, more flexible numbering). Amazon Connect after that for the AWS-native deployment, particularly relevant if you’re already running Connect for your support center. Both providers terminate calls into the same Henoyo container running in your cloud.
You choose. If your compliance policy requires call recording (financial services, healthcare), recordings stay with your telephony provider under your existing retention policy. Transcription happens inside your Henoyo container using Deepgram (or Amazon Transcribe for the AWS-native path); transcripts live in your audit trail with the same six-year retention as everything else.
Phone-number-to-user mapping is configured in the portal. Inbound calls match against that mapping; the audit trail captures the calling user. For unrecognized callers, Henoyo can either reject the call or route it to a fallback flow your team configures (for example: “please identify yourself by your Salesforce ID”).
Sub-2-second responses for short Prompts; 3–5 seconds for Prompts that fan out across multiple systems. The browser voice channel is at this latency today; telephony adds the PSTN round-trip but not the model round-trip.
Yes. Skills that write require explicit verbal confirmation: “Log this as a service visit on the Acme Northbrook account? Confirm or cancel.” Once confirmed, the Skill executes and the audit trail captures the field user, the field name, the before-value, and the after-value.
Get on the early-access list. We’ll walk through your Twilio or Amazon Connect setup, with your IT or telecom lead in the meeting if you want, show how the same Prompts you author once light up over a phone call, and answer the questions your compliance team is going to ask.