> For the complete documentation index, see [llms.txt](https://docs.bonzai.sh/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.bonzai.sh/readme.md).

# Welcome to BonzAI

BonzAI helps you write, create media and put AI agents to work on your own computer. Start with a useful result—a clearer email, an illustration, a narrated story or a repeatable business task—and build from there.

**New here?** [Choose your starting point](/start-here/getting-started.md), or follow [your first 10 minutes](/start-here/first-10-minutes.md). **Coming back?** See [what’s new](/start-here/whats-new.md).

## Choose the tool for your task

| I want to…                                      | Start here                                                             | What to expect                                                                      |
| ----------------------------------------------- | ---------------------------------------------------------------------- | ----------------------------------------------------------------------------------- |
| Try private text chat without installing an app | [BonzAI Web](/product-manuals/bonzai-web.md)                           | A model download in a compatible browser; prompts run on your device                |
| Make images, voices, music, video or a game     | [BonzAI Desktop](/product-manuals/bonzai-desktop.md)                   | A local studio with models selected for the task and your hardware                  |
| Give an agent a character and memory            | [Roleplay and companions](/product-manuals/bonzai-desktop/roleplay.md) | Local conversations and reusable context                                            |
| Turn a recurring task into a workflow           | [Smart Workflows](/product-manuals/bonzai-desktop/company-jobs.md)     | Teams, inputs, tools, activity and reviewable deliverables                          |
| Collect examples from webpages                  | [BonzAI+](/product-manuals/bonzai-plus.md)                             | A browser assistant with capture and dataset tools; check distribution availability |
| Publish work or understand the token economy    | [Web3 Guide](/ownership-and-network/guide.md)                          | Optional wallet actions, with prerequisites, costs and limitations explained        |

[Download BonzAI Desktop 4.1.0](/start-here/install-desktop.md) for Apple Silicon macOS, Windows x64, or Linux amd64 with CUDA or ROCm.

## Start free, keep control

Local chat, generation and local Smart Workflows do not require a wallet, a token balance or a subscription. You supply the hardware, storage and electricity. Models download separately, and larger models need more memory and time.

Local inference processes your prompts on your device. Internet access is still needed for downloads, updates and any online features you choose. Remote inference and connected tools send the inputs needed for that service; publishing can make selected files and metadata public. See [privacy and downloads](/start-here/privacy-wallets.md).

## Build on your work

A conversation can help shape an idea. Generation turns it into text or media. Datasets organize examples for training, while memory carries useful context into later tasks. Smart Workflows organize agents around a result you can inspect.

Saving locally is enough to use the output. Minting, staking, markets and paid remote services are optional next steps, not requirements for creation. An available screen does not prove a contract or market is live: use the [Web3 Guide](/ownership-and-network/guide.md) to understand the boundary before proceeding.

## Find your next step

* [Hardware and model choices](/start-here/system-requirements.md): choose something your computer can run comfortably.
* [Desktop navigation](/product-manuals/bonzai-desktop/navigation.md): find creation, agents, training and settings.
* [Themes](/product-manuals/bonzai-desktop/themes.md): choose a comfortable light or dark workspace.
* [Everyday workflows](/everyday-workflows/user-scenarios.md): turn a capability into a useful outcome.
* [Troubleshooting](/reference/troubleshooting.md): recover when downloads, models or generation fail.