Getting started
Create an account, whitelist the IP your switch sends from, point a trunk at the ingress, choose a plan and place a test call.
For The engineer connecting the first trunk.
Documentation
These guides cover everything between "we have a SIP trunk" and "we are reading a fraud verdict on every call": connecting your switch, the fields in an analysis, the REST API and the errors you can hit. Written for the engineer doing the work. Every value comes from the running system.
Start here takes you from nothing to a first analyzed call. Connect your switch has a step-by-step page for each platform we have written one for. Reference is where you go with a symptom, a questionnaire or an integration to build.
Create an account, whitelist the IP your switch sends from, point a trunk at the ingress, choose a plan and place a test call.
For The engineer connecting the first trunk.
Where to send the INVITE, transports and ports, IP authentication, codecs, DTMF, prefix routing and what we answer when something is wrong.
For Whoever owns the switch, SBC or firewall.
Verify that a route delivers your caller ID intact: pick a destination and a country, send a caller ID, read the verdict, the timing and the audio quality. Schedules and alerts.
For Wholesale and NOC teams who settle caller ID disputes with vendors.
Point a number you own at the platform, prove it with one call, and earn US$0.15 for every caller ID check that reaches it. Where to point it, the test, payouts, the API.
For Anyone with a working number in a country operators need to verify against.
What happens to a call after it ends, every field in the analysis output, and how to read a probability, a category and a recommendation.
For Fraud analysts and NOC staff reading verdicts.
Send calls from any plain-UDP switch, SBC or PBX to an OpenAI GPT-Live-1 or xAI voice agent: whitelist the switch, add a Voice AI destination, register the webhook, route a prefix.
For Operators putting an AI agent behind an existing switch.
Configure a chan_pjsip trunk on your Asterisk, whitelist its IP, and route every call, or one in ten, through the ingress with a short dialplan change.
For Asterisk administrators.
Configure a sofia external gateway on your FreeSWITCH, whitelist its IP, and bridge every call, or a weighted one-in-ten sample, through the ingress.
For FreeSWITCH administrators.
Configure your Kamailio to set $du towards the ingress and relay every initial INVITE, or a random share of them, through Open Voice Shield.
For Kamailio administrators.
Configure your OpenSIPS to set $du towards the ingress and relay every initial INVITE, or a deterministic sample, through Open Voice Shield.
For OpenSIPS administrators.
Build a Signaling Group towards the ingress on your SBC Edge, route calls to it from the Call Routing Table, and sample with a prefix subset.
For Ribbon SBC administrators.
Define Open Voice Shield as a vendor and SIP connection in your PortaSwitch, attach a vendor tariff, and include the routes in a routing plan.
For PortaSwitch administrators.
Create a Proxy Set, IP Group and IP Profile for the ingress on your Mediant, then send calls to it with an IP-to-IP Routing rule.
For AudioCodes SBC administrators.
Point an Origination URI or your TwiML at the ingress for inbound, send outbound through the relay to your Termination URI, and sample by weight or one if.
For Teams built on Twilio trunking or Programmable Voice.
Route numbers to an FQDN SIP Connection naming the ingress, send outbound through the relay to sip.telnyx.com, and sample in code or by number range.
For Teams built on Telnyx SIP Connections.
Add a Carrier whose outbound SIP gateway is the ingress, dial through it with a sip target, and whitelist your own server’s egress address.
For Teams running self-hosted Jambonz.
The two things every connection needs, an INVITE with SDP sent to a next hop, and symmetric RTP, plus the equivalents on SBCs and diallers.
For Anyone whose switch has no guide of its own.
Choose Economy, Power or Max, what each tier does, how switching behaves, and how the mode shows on the CDR and the ledger.
For Anyone deciding what a call analysis should cost.
Connect a channel in your own Slack workspace under Connectors, turn Slack on in an alert rule, and what the message carries. Add to Slack or a webhook URL, testing, throttling, and the API.
For Anyone who wants verdicts where the team already looks.
Where monthly invoices live, how they arrive by e-mail, what each statement shows, and how GST/HST applies for Canadian customers.
For Account owners handling billing and tax.
Recordings and captures are stored in Canada by default; choose the United States per account, for storage and for the media relay. What moves, what stays in Canada, the API.
For Account owners with a data residency requirement.
Authentication, the call, analysis, automation and billing endpoints, webhook payloads and signatures, errors and rate limits.
For Developers wiring Open Voice Shield into their own systems.
How an AI agent runs the platform: discovery files, the MCP server with OAuth sign-in, test calls a person confirms, verified numbers, and the REST flow with curl.
For Agent builders and the agents themselves.
The SIP reject codes and processing errors you can actually hit, what each one means, and the check that clears it.
For On-call engineers with a symptom in hand.
Authentication, tenant isolation, what is stored, retention and deletion, transport and the controls around recordings.
For Security reviewers and procurement.
| If you are… | Start with | Then read |
|---|---|---|
| Connecting the first trunk | Getting started | SIP integration |
| Configuring a switch, SBC or firewall | SIP integration | Troubleshooting |
| Reading verdicts as an analyst | Concepts | Read the verdict |
| Building an integration | API reference | The interactive API explorer |
| Chasing a call that failed | Troubleshooting | SIP integration |
| Reading invoices and tax | Invoices | AI modes |
| Getting alerts where the team looks | Slack alerts | The default alert rule |
| Letting an AI agent run it | For AI agents | API reference |
| Reviewing us before you buy | Security | Talk to us |
Open Voice Shield is an extra SIP hop between your switch and wherever the call was already going. It authorizes the call by source IP, relays the signalling and the media, records both directions, and routes the call onward to a destination you configure. After the call ends it transcribes the recording, analyzes it and attaches a fraud probability, a category and a block, review or allow recommendation to the call record.
Longer version, with the diagram: how a call travels through Open Voice Shield.