> For the complete documentation index, see [llms.txt](https://trustlevel.gitbook.io/knowledge-base/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://trustlevel.gitbook.io/knowledge-base/what-is-trustlevel.md).

# What is TrustLevel?

TrustLevel builds tools and infrastructure that make **trust measurable and verifiable**. In a world increasingly shaped by AI — where information is abundant but reliability is scarce — knowing *what* and *whom* to trust has become the hard part.

## Our approach

We don't ask whether something is *true*. We ask **how much it can be trusted** — and we treat that as a probability, not a yes-or-no verdict. Any claim, review, or AI output can be examined along four dimensions:

* **How confident is it?** — Trust is a probability, never a binary.
* **What is it based on?** — We trace the evidence and reasoning behind a claim.
* **Who disagrees?** — We surface where sources contradict one another.
* **How fragile is it?** — We test how conclusions hold when assumptions change.

## Where the method comes from

This isn't theory. It grew out of years of hands-on work in **Web3 grant programs** — evaluating hundreds of funding proposals in Project Catalyst, SingularityNET's Deep Funding, and beyond.

That work taught us a practical lesson: the quality of an evaluation depends on **who** is doing it and **how much their judgment has proven reliable over time**. In other words, trust can be modelled as **Reputation × Expertise** — measured through structured peer review rather than assumed.

## From method to product

That principle is exactly what **REX** puts to work: a peer-review system that scores funding proposals by combining reviewer expertise with earned reputation. It's the most direct expression of our methodology — see [REX](/knowledge-base/products/rex.md).

Alongside it, the **ZK Voting App** applies the same commitment to verifiable trust in a different domain: anonymous, tamper-proof voting on Cardano — see [ZK Voting App](/knowledge-base/products/zk-voting-app.md).

The thesis is simple: **trust should be measurable, not assumed.** Every product puts that principle to work in a different domain.