> For the complete documentation index, see [llms.txt](https://astrai-1.gitbook.io/astrai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://astrai-1.gitbook.io/astrai/vision-and-pain-points.md).

# Vision & Pain Points

**ASTRAI** is a decentralized protocol for on-chain AI compute and model collaboration, powered by crypto-native incentives and verifiable infrastructure. It provides an open and censorship-resistant alternative to cloud monopolies, enabling:

* Distributed GPU/ASIC compute markets
* On-chain model training with contributor rewards
* Tokenized AI models with programmable royalties
* Transparent, DAO-governed infrastructure for AI

> **We believe the future of AI should be open, permissionless, and owned by its contributors.**

## **Problem Statement**

#### Centralized AI suffers from:

1. **Compute concentration**

   > AI development is restricted to those who can afford large cloud infrastructure (Big Tech), leaving independent devs and DAOs out.
2. **No creator ownership**

   > Model developers, fine-tuners, and data providers can’t claim or monetize ownership of the models they contribute to.
3. **No training incentives**

   > There is no way for regular users to earn rewards by improving, customizing, or training AI models.