> ## Documentation Index
> Fetch the complete documentation index at: https://docs.upliftai.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

Uplift AI lets you **build** and **deploy** a **voice agent** in a minute.

Deploy voice agents on your website, dial outbound calls to your customers, and have them answer the phone.

Experience a call from our voice agents [here](https://upliftai.org/#calling-demo).

## What people build (and what it costs)

Voice agents at a price that works for Pakistan: **12 PKR / min** of conversation, with a **Pakistani phone number**.

Currently, the best use cases we have seen:

* [**Order confirmation (COD)**](https://upliftai.org/industries/e-commerce)
* [**Delivery address confirmation & rescheduling**](https://upliftai.org/industries/couriers-delivery)
* [**Appointment booking & no-show reduction**](https://upliftai.org/industries/healthcare)
* [**Lead qualification**](https://upliftai.org/industries/real-estate) — i.e. invitation to events
* [**Payment reminders**](https://upliftai.org/industries/financial-services)
* [**Routine support calls**](https://upliftai.org/industries/telecom)

But the agents are only limited by your imagination. You can share your ideas and work with us at [founders@upliftai.org](mailto:founders@upliftai.org)

## How can you get started

Create your agents on the [portal](https://upliftai.org/app/calling) or directly via the [API](/api-reference/assistants/create-an-assistant).

For proper details I recommend reading [Quick Start](/voice-agents/getting-started/quickstart-phone) but for a one liner:

Just share details about your assistant e.g. *'You are an assistant by company X calling farmers to confirm if they are interested in a field visit....'* in the prompt, and launch your assistant.

## Tech details for techies

**You configure** and **choose** the voice components:

* **Speech to Text**: choose across leading providers like Soniox, or OSS models like Whisper
* **LLM**: All major providers supported: OpenAI, Gemini, Groq, Cerebras. Try closed and open-source models.
* **Text to Speech**: Choose among 80+ Uplift AI TTS voices, or another provider (*we want to know why!*).

<Note>
  Our TTS requires Urdu in Nastaliq script, not Roman Urdu, for best results. This is a simple LLM prompt change for output style.

  Example prompt addition:

  ```
  Only respond in Nastaliq Urdu

  - Good: "آج موسم بہت خوشگوار ہے۔"
  - Bad: "Aaj mosam bohat khushgawar hai."
  ```
</Note>

Our agents run on LiveKit, deployed **in Pakistan**. This allows us to do calling with Pakistani phone numbers, and some of the lowest end-to-end latencies.

Anything compatible with LiveKit for UI — components, application libraries — works on top of us.

**Our APIs** and UI **allow** the ability to:

* **schedule calls**, call customers within business hours automatically — see [Campaigns](/voice-agents/campaigns/getting-started)
* hook up your **webhooks**:
  * receive **call analysis** and **transcripts** as calls finish — [Webhooks](/voice-agents/webhooks/overview)
  * **tool calls**: we can call your service DURING the call — [Webhook tools](/voice-agents/tools/webhook-tools)
* Call **personalization** — [per-call instructions](/voice-agents/personalization/per-call-instructions):
  * Pass call specific details that append on to your main agent prompt
    * Example: for a lead qualification call to a farmer, pass details the prompt can utilize, like crop name, disease, applicable products and benefits.
* Drop contacts directly from a **Google sheet** or **file** — [Contacts](/api-reference/contacts/upload-a-contact-file).

#### Architecture Overview

Web and phone calls are seamless from an agent configuration point of view — we take care of the calling details under the hood.

##### **Phone calls**

```mermaid theme={null}
graph LR
    A[Caller] <-->|phone call| B[Agent]
    B --> C[LLM]
    B --> D[STT]
    B --> E[TTS]
    B -.->|custom tools| F[Your services]
```

##### **Web calls**

In web calls, you can provide / update tools and instructions on the fly. This lets you give your agent the ability to control the UI or provide information to the agent without having to do custom integrations.

```mermaid theme={null}
graph LR
    A[Browser] <-->|WebRTC| B[Agent]
    B --> C[LLM]
    B --> D[STT]
    B --> E[TTS]
    B -.->|tool invocation| A
    B -.->|custom tools| F[Your services]
    A -->|update instructions / tools| B
```