> 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/products/rex.md).

# REX

REX is TrustLevel's peer-review system for **funding decisions**. It helps grant programs, DAOs and foundations evaluate proposals in a way that is structured, transparent and hard to game — by weighting each review according to the reviewer's **Reputation × Expertise** (hence the name).

REX grew directly out of our work redesigning the proposal-review process for **Project Catalyst** (Cardano's community funding program). It is a **custom product**, not a self-serve app: it is deployed and operated for a specific program. For Catalyst it runs in production; for other programs it is available on request.

## Why it exists

Open, community-driven funding has a recurring problem: reviews vary wildly in quality, expertise rarely reaches the proposals that need it, and there's little accountability for a careless or biased assessment. REX addresses this by making review quality itself measurable — and by letting a reviewer's track record shape how much their voice counts.

## How it works

**Two-stage review.** Every proposal goes through two layers:

1. **Proposal reviews** — reviewers assess a proposal against a structured framework.
2. **Peer reviews** — other reviewers evaluate the reviews themselves, which is how review quality is scored and reputation is earned.

**Six evaluation criteria.** Each proposal is rated across Relevance, Innovation, Impact, Feasibility, Team and Budget, with descriptive anchors for each rating. A quick *temperature check* up front decides whether a full review is warranted.

**Expertise-based matching.** Proposals are routed to reviewers by a tiered match — technical/community/product expertise first, shared interests second, random assignment only as a fallback. Conflicts of interest are excluded automatically.

**Reputation.** Peer ratings feed a reviewer's reputation over time, which in turn influences matching and standing. Good, specific, insightful reviewing is rewarded; low-effort reviewing is surfaced and down-weighted.

**AI assistance.** Language-model summaries condense long proposals so reviewers can focus their judgment where it matters.

## Who it's for

Grant programs, DAOs, foundations and any funding body that runs an open or semi-open review round and wants more reliable, accountable outcomes than unstructured community voting provides.

## Learn more

Source code and full technical documentation: [github.com/TrustLevel/Grant\_Review\_Tool](https://github.com/TrustLevel/Grant_Review_Tool)