> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.voicebip.com/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.voicebip.com/_mcp/server. # Use Cases > End-to-end recipes that combine Voicebip primitives into a working thing — an answered support line, an outbound reminder campaign, a WhatsApp agent. Each page is copy-paste runnable against the sandbox. The [Guides](/guides/agents) explain one primitive at a time. **Use Cases** put them together: each page builds a complete, working scenario you can run end-to-end and adapt to your own product. Every page follows the same shape, so once you've done one you know how to read them all: 1. **What you build** — one paragraph and a diagram of the finished thing. 2. **Prerequisites** — the API key, number, or webhook receiver you need first. 3. **Steps** — numbered, each with a runnable `curl` (or SDK) call and the response you should see. 4. **Config** — the full agent / campaign / tool JSON in one block, ready to copy. 5. **Test it** — how to trigger the flow in [Sandbox Mode](/reference/sandbox-mode) (`pk_test_`) with **no real calls, no billing**. 6. **Next steps** — what to layer on, and links to the deeper Guides. > **Note** > > Everything here works against the sandbox. Use a `pk_test_` key to walk through a use case end-to-end — synthetic call/message lifecycle events fire with real HMAC signatures, and billing is ₦0. Swap to your `pk_live_` key when you're ready to go live. ## Start here (must-have set) These five cover the platform's core surfaces. Do them in order the first time. | Use case | You build | Primitives | | ------------------------------------------------------------------------ | ------------------------------------------------------------ | -------------------------------------------- | | [Inbound support agent + Knowledge Base](/use-cases/inbound-support-kb) | A +234 line that answers callers from your own docs | Numbers · hosted AI · KB/RAG · webhooks | | [Voice agent that takes actions (tool-calling)](/use-cases/tool-calling) | An agent that calls *your* API mid-call ("track my order") | Tool definitions · `tool.invocation` webhook | | [WhatsApp AI agent](/use-cases/whatsapp-agent) | A WhatsApp number that holds a real support conversation | Messaging · conversations · hosted AI | | [Outbound reminder campaign](/use-cases/reminder-campaign) | A list-driven campaign that confirms appointments / payments | Campaigns · voice/SMS · delivery webhooks | | [Verification call-back (voice OTP)](/use-cases/verification-callback) | An outbound call that reads a one-time code | Outbound calls · dynamic greeting | ## Go deeper Primitive-focused reference flows and advanced combinations. | Use case | You build | | ---------------------------------------------------------------- | ----------------------------------------------------------------- | | [Knowledge Base deep-dive](/use-cases/knowledge-base-deep-dive) | RAG search *without* a call — ingest, poll, query, BYOM retrieval | | [Campaigns deep-dive](/use-cases/campaigns-deep-dive) | Full campaign lifecycle, per-recipient params, retry & counters | | [Multi-agent routing](/use-cases/multi-agent-routing) | One number → sales vs. support agents by rule | | [Bring your own model (BYOM)](/use-cases/byom-custom-llm) | Voicebip handles telephony; your endpoint runs the LLM | | [Multi-channel cascade](/use-cases/multi-channel-cascade) | Reach a contact by voice → SMS → WhatsApp, automatically | | [Browser voice widget](/use-cases/browser-widget) | A "talk to our agent" button on a web page (WebRTC) | | [Drive Voicebip from an AI assistant (MCP)](/use-cases/mcp-tour) | Provision numbers and launch campaigns conversationally | ## What Voicebip doesn't do (yet) We'd rather you know the edges up front than discover them mid-build. If your use case needs one of these, here's the honest state and the workaround: | You might want… | Today | Workaround | | -------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | | **Structured data capture** (typed JSON extracted from a call — survey answers, lead fields) | No native structured-output schema on agents | Have the agent record answers via a [tool](/use-cases/tool-calling), or parse the `call.completed` transcript yourself | | **First-party calendar/booking** | No built-in Cal.com/Google Calendar integration | Wire your own booking API as a [tool](/use-cases/tool-calling) | | **Visual flow / pathway builder** | Routing is code-first rules, not a drag-and-drop canvas | Model each branch as a rule → dedicated agent ([Multi-agent routing](/use-cases/multi-agent-routing)) | | **Context-preserving agent-to-agent transfer ("squads")** | `transfer_call` moves the call; cross-agent memory hand-off isn't guaranteed | Keep shared context in your backend, keyed by `call_id` | A few things exist as **primitives but don't yet have their own cookbook page** (they're on the roadmap): streaming a live transcript to your frontend ([SSE reference](/reference/streaming)), recording playback ([guide](/guides/recordings)), and webhook idempotency/dedup ([best practices](/reference/best-practices)). The building blocks are all shipped — only the worked example is pending. ## A note on Nigeria Most of these examples use Nigerian numbers (`+234…`), the `Africa/Lagos` timezone, and CBN call-window defaults (08:00–20:00) because that's where the carrier infrastructure lives. Agents also speak **English, Hausa, Yoruba, Igbo, and Nigerian Pidgin** — set `language` on the agent (`en` · `ha` · `yo` · `ig` · `pcm` · `auto`). Where a use case touches recording or opt-out, it follows [NDPR](/guides/contact-erasure) defaults.