Programmable voice is a set of application programming interfaces (APIs) or software that let applications place, receive, route and control phone calls in code. It is most commonly delivered as a cloud service. Instead of configuring a fixed phone system, developers write code, or use a visual flow builder, that decides what happens on each call: play a message, collect keypad input, connect to an agent, record, transfer or start a conference. Depending on the deployment, the provider supplies the connection to the public telephone network (PSTN) or SIP connectivity itself, or integrates with a carrier you bring, a model often called bring your own carrier (BYOC). Cloud services are usually priced mainly per minute. Programmable voice is one of the core building blocks of a communications platform as a service.
At a glance
- Programmable voice gives software control over phone calls through APIs, usually as part of a CPaaS platform.
- Common capabilities include outbound and inbound calling, call routing, recording, conferencing and keypad or speech input.
- Your application typically tells the platform what to do on each call when the platform sends it an event, often as a webhook.
- Depending on the deployment, the provider supplies PSTN or SIP connectivity, or integrates with your own carrier (BYOC); many platforms also connect to browsers and apps.
- Cloud services are priced mostly per minute, with feature, number and country costs that vary by provider.
What problem it solves
A traditional phone system or contact center platform does what its menus allow. When a business needs calling built into its own process, such as a delivery app that connects customers to couriers, a clinic that calls patients with appointment confirmations or a support line that routes callers based on account data, fixed menus often fall short.
Programmable voice lets a development team make calling part of the application. Calls can be triggered by business events, routed using data from a CRM or database, and logged back to the systems that need them. With a cloud service, the provider usually runs the media servers and capacity and either supplies carrier connections and phone numbers or integrates with a carrier you bring, so the team doesn’t have to buy or run telephony equipment.
How it works
Numbers and connectivity. Depending on the deployment, you rent phone numbers and public telephone network (PSTN) connectivity from the provider, or keep your own carrier and numbers under a bring your own carrier (BYOC) model. Many platforms connect to your existing carrier over SIP, using SIP trunking, and to browsers or mobile apps through WebRTC.
Call control. When a call arrives, the platform notifies your application, commonly with a webhook to an address you set. Your application replies with instructions, in a markup language or JSON format that varies by provider: say a message, gather digits, dial another number, join a conference or record. Some platforms instead stream call events and let your application send commands back as the call progresses. Outbound calls start with an API request.
Building blocks. Common features include text-to-speech, speech recognition, keypad input for interactive voice response (IVR) menus, call recording, transcription, conferencing, call queues and transfers. Many providers also stream call audio to your application for real-time processing such as voice AI. Which features are included, and how mature each is, varies by provider.
Events and records. The platform reports call status, duration, recordings and errors back to your systems, usually through webhooks and logs, so calls can be tied to customer records and billing.
Patterns built on it. Common uses include notifications and reminders, click-to-call, call tracking, number masking, voice verification codes and custom IVRs.
For buying guidance, see our communications platform as a service solution page.
When it matters for buyers
- Calling needs to live inside your own software, triggered by business events or using your data for routing.
- Custom call flows don’t fit a standard phone system or a full contact center as a service (CCaaS) platform, or would be expensive to add there.
- Call volumes are growing, and per-minute rates, feature charges and concurrency limits start to matter.
- You already have a carrier and want to keep its rates and numbers while adding programmable call control.
- Regulatory obligations apply. Emergency calling, call recording consent, caller ID rules and automated calling rules all vary by country and, in some cases, by state. The platform carries out your instructions; compliance stays your responsibility, so involve counsel.
Questions to ask vendors
- What are your per-minute rates for our call types and countries, and which features cost extra?
- How does call control work: webhooks with markup, a call-control API, or both?
- What concurrency limits and rate limits apply to our account?
- Can we bring our own carrier or SIP trunks, and keep our existing numbers?
- How do you handle emergency calling on numbers we use with your platform?
- What is your API uptime SLA, and what happens to calls in progress if our application or webhook is unreachable?
- Which countries can we get numbers in, and what registration or documentation is required?
How it differs from SIP trunking
SIP trunking connects a phone system you already run, such as an on-premises PBX or Microsoft Teams, to the public telephone network. It carries calls, but the phone system decides what to do with them. Programmable voice adds the call-control layer: your software decides what happens on each call through APIs. Many programmable voice providers also sell SIP trunking, and many let you combine the two, using your own trunks for connectivity and their APIs for call logic.