Too much, not too little
Hundreds of launches a week, near enough identical at a glance. Volume is not a shortlist, and a feed is not research.
Private beta · 2026
A faster way to discover
and evaluate web3 ventures.
Every venture we publish has been gone through by hand: who the team really is, what they shipped before, whether the audience is people or bots. Each one scored against the same rubric, built only from evidence anyone can go and check. Curated, not a feed.
The premise
Most of what matters is public.
Almost none of it has been checked.
Claims without evidence
Track records travel much further than the proof behind them. A history nobody went and verified is a story told well.
The good ones stay quiet
The most serious teams are often the hardest to find early, because they are busy building rather than posting about it.
Method
Researched, scored, published.
Every venture is gone through by hand.
Not a filter sitting over a firehose. Anything that makes it in has been deep searched: the people behind it traced back through what they have built before, and the on-chain record read rather than the pitch.
- Team traced back through their earlier work
- Prior launches found, including the ones that failed
- On-chain history read directly, not taken on trust
One rubric, and every input is public.
The same criteria are applied to every venture, so that two of them can actually be compared. Each input traces back to a source you can open yourself, which is what makes a score worth arguing with.
- Identical criteria applied to everything covered
- Each input traceable to a public source
- Thin evidence lowers a score rather than quietly going missing
A written report, not a rating.
The number is the summary, not the work. What gets published is the reasoning, the evidence and the parts that did not add up, revised when the facts move. Ventures we cover get it directly too, as advice.
- The reasoning published, not just a verdict
- Revised when the underlying facts change
- Direct guidance for the ventures we cover
Panels are illustrative.
What we score
The questions worth answering properly.
Team track record
Who they are, what they built before this, and how those ended.
Identity
Doxxed, pseudonymous with a traceable history, or fully anonymous. Stated plainly, never insinuated.
Launch history
How many they have shipped, how many are still alive, and what happened to the rest.
On-chain metrics
Holders, liquidity, treasury and flows, read off the chain rather than off a dashboard.
Genuine engagement
Whether the audience is actually people. Bought followers and botted replies are measurable.
Public sources
Every figure carries the source it came from, so a score can be checked instead of trusted.
The most ambitious teams rarely have the loudest launch. Finding them is the work. Standing it up with real research is the point.
Dalmanutha exists to put the promising ones in front of people, with the background to stand behind them, and to help them grow from there.
Waitlist
Early access opens in cohorts.
Leave an address for the first reports. Building something you want looked at properly? Use the same form and say so. No newsletter, no forwarding, one message when it is your turn.