Regarding
$DESC, the product may launch today.
I did some research, and here’s everything you need to know:
Supported by:
AR, Molecule , BankrBot, Akash Network
1Claw AI has already been successfully integrated into DescAI.
Team Lead Coby recently participated in the Base hackathon.
I believe Base will support a project that has been incubated within its ecosystem.
The core idea behind DescAI:
DeScAI is a project at the intersection of DeSci (decentralized science) and AI. Its core, Agent-Core, is essentially an "automated scientific review factory": an autonomous AI agent that finds scientific content across crypto-science ecosystems on its own, runs it through a pipeline of language models, and produces a structured quality assessment.
Crawling. The agent gathers source data from three places: ResearchHub (scientific papers and funding proposals), Molecule IPNFTs (tokenized intellectual property from research DAOs), and Pump Science (chemical compound tokens for longevity research). github
Reviewing. Each content type has its own LLM pipeline. For example, the articles pipeline is a 13-step process: extracting scientific claims from a PDF, routing them, and grading the empirical evidence, including originality checks against the OpenAlex database. github
Output. Every run produces a standard bundle: review.json with integer scores from 0 to 100, overview.json — a plain-language summary, and evidence_audit.md — a provenance audit trail showing the sources behind each conclusion. github
Publishing. Finished reviews can be published to Arweave (a permanent data storage blockchain) and backed up to private Cloudflare R2 storage. Writing to Arweave makes a review permanent, immutable, and publicly verifiable. github
In short: it's an AI reviewer that automatically checks the quality of science in crypto-science projects and records its verdicts on the blockchain.
Where it will be applied
The project addresses the main pain point of the DeSci ecosystem: there are plenty of tokenized "science" assets, but almost no independent expert evaluation. Concrete use cases:
Due diligence for DeSci token investors. On Pump Science, people trade chemical compound tokens (like RIF and URO) tied to real longevity experiments. The agent provides an independent AI assessment of a compound's scientific merit before someone buys the token. Gate LearnThe Defiant
Evaluating funding proposals. ResearchHub collects crowdfunded research proposals — the agent reviews them and helps the community decide what to fund.
Screening research DAOs. The DAO pipeline takes an IPNFT "dataroom" from Molecule and produces a six-category review — in other words, it evaluates tokenized scientific projects and their intellectual property. github
Replacing/supplementing traditional peer review. Conventional peer review is slow and closed; here, a review is generated automatically, comes with an evidence trail, and is stored publicly and permanently.