Napster API

Video AI agents
with a memory.

Agents your users can see, talk to, and call, and that recognize them when they come back. $3.50 an hour, everything included.

Read the Docs

Your first 100 minutes are on us. No card required. Also available on Microsoft Marketplace.

Thousands

Pre-built personas

~300ms

Avg. latency

$3.50

Per hour, all in

SOC 2

Type II certified

Quickstart

One prompt. Two steps.
One AI agent.

One prompt, pasted into the AI coding tool you already use. First, agentify your website so an agent can see and operate it. Then install an AI agent that can see, speak, remember, and take action.

Agentify your website

Prepare your website for an AI agent to operate it, not just understand it. The agent can see what’s happening on the page, understand what the visitor is doing, and take action across your site. Search products, update a cart, complete a booking, navigate between pages, using the same capabilities that power your website today. Your website becomes a place where an AI agent can actually get things done.

Install the AI agent

Now bring your agent to your website. Give it a face, a voice, your product knowledge, and the memory to recognize returning visitors. Your website is just the beginning: once your agent is up and running, extend the same experience to the phone, kiosks, and wherever your customers are.

Read the Docs

Works with Claude Code, Cursor, Codex and any tool that supports Agent Skills. You’ll need access to your site’s code.

Developer-first APIs

Clean REST APIs and a Web SDK that works across all major web frameworks. Mount an agent in any container element.

Real-time streaming

WebRTC for video and audio, WebSockets for voice-only. Low-latency, adaptive quality, no infrastructure for you to manage.

Agent orchestration

Napster coordinates the model, avatar rendering, voice synthesis, and tool execution, delivering a unified experience to your users.

Channels

One agent,
every channel.

Deploy the same agent across video, audio, and telephony. Same identity, same knowledge, same behavior, regardless of how your users connect.

Video

Real-time video avatars with synchronized lip-sync and expressive gestures. Embed directly in your web application via our Web SDK.

Avatar Library

Custom Avatars

Audio

Voice-only agents for lightweight integrations and audio-first experiences. Low bandwidth, low latency.

Scalable

Low Latency

Connect your agents to phone lines. Handle inbound/outbound calls with the same conversational intelligence.

VoIP

SIP

Platform

Everything you need to
ship production agents.

Napster is a full agent platform. Define your agent, equip it with knowledge, deploy it, and monitor it, all from one place.

Agent Builder

Define your agent's identity, behavior, and system instructions. Choose an avatar or create a custom one from a single image.

Knowledge Base

Upload documents, websites, and structured data. Your agent retrieves the right context at conversation time.

FAQ Collections

Define question-answer pairs for deterministic, always-correct responses on critical topics.

Agentic Tools

Equip your agent with skills, invoke external APIs, trigger workflows, and take action in real-time during conversations.

Memory

Agents remember context across sessions. Previous interactions inform future conversations for personalized experiences.

Safety & Governance

Content moderation, crisis detection, and full audit trails, always on, not optional per agent.

Bring Your Own Model

Connect your own AI deployment. Inference runs on your models, in your tenant, under your own provider agreement.

Thousands of Personas

Pre-built characters ready to go. Or create a digital twin from a single photo.

Drop-in Web SDK

JavaScript SDK for React, Vue, Angular, or vanilla JS. Mount an agent in any container element with full docs and examples.

How it works

Build. Deploy.
Monitor.

Three steps from idea to production agent. The platform handles the complexity so you can focus on the experience.

Build your agent.

Define your agent’s identity and capabilities through the dashboard or API. Shape personality with our Large Persona Model, 30+ psychometric traits you can fine-tune. Add knowledge sources, configure FAQs, and set up tool integrations.

System Instructions

FAQs

Custom Avatar

Knowledge Base

Tool Calling

Deploy everywhere.

Embed video agents on any web page with our drop-in Web SDK via WebRTC. For audio-only agents, connect over WebSockets. VoIP and SIP for telephony.

Web SDK

WebRTC

WebSockets

VoIP & SIP

Monitor & iterate.

Review session transcripts to understand how your agents perform. Identify gaps, refine knowledge, and improve your agent continuously.

Session Transcripts

Continuous Improvement

pricing

One platform.
Two ways to pay for the model.

Let Napster bring everything, or bring your own model and run it on our agentic stack. The presence, memory, safety layer, tools, and SDKs are the same either way.

what’s included

No setup fees or minimum commitment

All modalities included - video, audio, telephony

Unlimited agents and knowledge bases

Full API and SDK access

Safety and governance included

What does that look like?

A 5-min customer call

$0.05

100 calls / day

~$5/day

A full-time agent, 8h/day

$4.80/day

10 agents running all month

~$4,320/mo

Everything included

Napster brings everything

$3.50/ hour

$0.058 per minute

Our agentic stack and our model. Nothing to deploy and no model bill to track.

This is the whole price. There is no second invoice.

Bring your own model

$0.60/hour

$0.01 per minute, plus what your provider charges you

Run your own foundation model on our agentic stack. We handle the presence and the orchestration, you supply the model.

Your model provider bills you directly. Bring your own agent: coming soon.

Included in both

✓ Video, audio, and telephony

✓ Knowledge bases and FAQ collections

✓ Safety and governance, built in

✓ SOC 2 Type II certified

✓ Memory across sessions

✓ Agentic tools

✓ Web SDK, WebSockets, and REST API

✓ No setup fees or minimum commitment

What it costs in practice

Scenario	Napster brings everything	Bring your own model
A 5-minute conversation	$0.29	$0.05
One agent, 8 hours a day	$28 a day	$4.80 a day
A support line, 2,000 conversations a month	about $580 a month	about $100 a month

Conversations are costed at five minutes. Bring-your-own-model figures exclude what your own provider charges you.

Start building

Get it on Microsoft Marketplace

Your first 100 minutes are on us. No card required to start.

Frequently asked questions

What is the difference between bringing my own model and letting Napster bring everything?

Napster brings everything means the model is included, so there is nothing to deploy and one rate covers the whole thing. Bring your own model means you run your own foundation model on our agentic stack: we handle the presence and the orchestration, you supply the model and pay your provider directly. Support for bringing your own agent, connecting an agent you already run elsewhere, is coming soon.

Which one is cheaper?

It depends on your model. At $0.60 an hour plus your own inference, bringing your own model is usually cheaper at volume, and it is the right choice if you already run a deployment or need inference to stay in your tenant. If you are starting from nothing, $3.50 an hour is the complete number and there is no second bill to forecast.

How is usage measured?

Usage is billed by the minute, rounded up per session. A 15-second session bills as one minute; a session of 3 minutes 35 seconds bills as 4 minutes. The minimum charge for any session is one minute.

Is there a free allowance?

Yes. Every new account starts with 100 minutes of agent time, and you do not need a card to begin. Add a card only when your minutes run out.

Are there volume discounts?

The API is self-serve by default. Committed build-with deals can be sales-assisted, so talk to us if you are planning at volume.

Can I use it without a commitment?

Yes. No setup fee and no minimum.

Should I buy direct or through Microsoft Marketplace?

Buy direct if you want to be building in a few minutes on a card. Buy through Marketplace if your organisation already has an Azure commitment and you would rather draw this against it, or if Microsoft procurement is how you buy software. Same platform and same rates either way.

If I buy through Microsoft Marketplace and directly, is that one balance or two?

They are separate. Buying direct is prepaid: you top up credits and your usage draws them down. Buying through Microsoft Marketplace is postpaid: you are billed at the end of the billing period for what you used, and if you used nothing you pay $0. Same rates either way.

Do I need my own LLM?

Only if you bring your own model. If you pick Napster brings everything, the model is included and there is nothing to deploy. When you do bring your own, model costs are billed by your provider: an Azure deployment, for example, is billed through your Azure subscription.

Where are the docs?

Full API reference, SDK guides, and examples are at developers.napster.com.

Security

Your data
stays yours.

Enterprise-grade security and compliance, built in from day one.

Bring Your Own Model

Connect your AI deployment. Inference stays in your tenant on your models. Napster orchestrates the session and holds only the conversation state you configure: memory, transcripts, and knowledge.

SOC 2 Type II

Independently audited annually. Enterprise-grade security, governance, and data management.

GDPR & US Compliance

Compliant with European and US state privacy laws, including CCPA. Data residency controls included.

Encryption at Rest

All stored data is encrypted at rest using AES-256. Keys managed through industry-standard KMS.

get started

Start building
with Napster.

Ship your first AI agent today. No setup fees, no commitments.

Get it on Microsoft Marketplace

Your first 100 minutes are on us.

Video AI agents with a memory.

One prompt. Two steps.One AI agent.

One agent, every channel.

Everything you need to ship production agents.

Build. Deploy. Monitor.