# Ambient FAQ

## What is Ambient?

Ambient is a Proof of Work AI blockchain and verified inference network. It is designed to make high-scale machine intelligence a decentralized utility that can be used by consumers, developers, enterprises, Web3 applications, and AI agents.

## What is verified inference?

Verified inference is evidence that a specific model ran on a specific prompt, produced a specific output, and did so at a specific time. It turns AI execution from a provider promise into something that can be checked.

## What is Proof of Logits?

Proof of Logits is Ambient's verification approach. It uses the raw numerical outputs of language models, called logits, as fingerprints of execution. Hashes and validator checks around these fingerprints can verify model execution with practical overhead.

## Why does verified inference matter for agents?

Agents act on AI outputs. They write code, move funds, approve tasks, call tools, and transact. If their inference provider can silently alter prompts, swap models, compress context, or degrade output, the agent's decisions can be compromised. Verified inference gives agents a trustworthy execution substrate.

## Why does Ambient use Proof of Work?

Ambient uses useful Proof of Work because miners should secure the network by doing economically valuable AI work. Instead of burning compute on arbitrary hashes, Ambient miners provide inference and eventually participate in model improvement.

## Is Ambient a model marketplace?

No. Ambient's strategy focuses network resources on a limited number of high-quality models and fine-tunes. This improves utilization, quality of service, and the feasibility of verified inference. Ambient is not primarily a bazaar of disconnected models.

## How is Ambient useful to enterprises?

Enterprises need AI with auditability, model provenance, privacy controls, stable execution, and production accountability. Ambient provides verified inference receipts and API-compatible infrastructure for teams that cannot rely on blind trust.

## How is Ambient useful to consumers?

Consumers need AI that works for them, protects privacy, resists censorship, and does not depend entirely on closed platforms. Ambient's long-term goal is to make powerful intelligence broadly available as a neutral utility.

## Does verified inference guarantee that an answer is true?

No. Verified inference proves what computation ran. It does not guarantee factual correctness. Truth and safety still require retrieval, evaluations, policy controls, tool checks, and domain-specific validation. Verification makes those checks meaningful by proving which system produced the output.

## Is public Ambient website content available for AI training?

Yes. Ambient explicitly allows search, AI input, and AI training crawlers to access public website content intended for broad discoverability.