LiveBTCloadingETH...SOL...XRP...Fear and Greed ...
Playbooks

A 30-minute AI research workflow for vetting a token

AI assistants are good at the boring parts of research: reading, summarising, comparing. A repeatable workflow, the prompts to use, and the checks you should never hand over.

01sources02summary03tokenomics04red team05verify3 hours of research, done in 30 minutes
Short version

Feed the AI primary sources, demand quotes for every claim, use it to red-team, and verify ownership and yield source yourself.

AI assistants have changed how I research crypto projects. Not because they know the answers, but because they are excellent at the boring parts: reading, summarising, comparing. Here is the workflow I use, end to end.

#Step 1: collect primary sources first

Before any AI, I gather the raw material myself: the official docs, the token contract page on a block explorer, the audit reports, the team's own announcements, and the governance forum. AI is much more reliable when you hand it sources than when you ask it to remember.

#Step 2: structured summary

I paste the docs and ask:

Summarise this protocol in a table: what it does, who pays fees, where the yield comes from, what can go wrong, and what the token is used for. Quote the line from the document for each answer.

Asking for quotes is the single best trick. It makes fabricated answers easy to spot.

#Step 3: tokenomics sanity check

From this supply schedule, list every unlock in the next 12 months as a percentage of circulating supply. Flag any month above 5%.

Then I verify the biggest numbers on the explorer myself.

#Step 4: red team it

Act as a skeptical analyst. Give me the five strongest reasons this project could fail, ranked by likelihood, citing the documents.

This is where AI earns its keep. It surfaces risks I would skim past.

#Step 5: the checks I never delegate

  • Contract ownership and upgrade keys: checked on the explorer, by me.
  • Where the yield really comes from: if I cannot explain it in one sentence, I pass.
  • Team and treasury wallets: I look at actual movements.

#Where AI still gets it wrong

  • Dates and numbers from memory are often stale or invented.
  • It is too polite about weak projects unless you ask it to be critical.
  • It cannot tell you whether the market will care.

What to keep for yourself

  • Feed AI primary sources and demand quotes.
  • Use it to summarise, compare and red team, not to decide.
  • Verify ownership, yield source and wallets yourself.

Brian Ho

Founder

Brian has traded crypto and FX since 2017, lost money to two martingale bots, and now tests AI trading agents on small accounts before writing about them.

Changes to this article

Oct 14, 2026: reviewed and updated. First published Sep 8, 2026. Spotted a mistake? Tell us, and we will fix it and note it here.

This is education and personal experience, not financial advice. Crypto and leveraged trading can lose you money fast. We may earn from sponsors shown on this page; they never edit our conclusions. How we make money.