DARWA™ is live. Type a name. Watch a brand exist — in seconds. No prompt. No templates. No code. No design decisions. The domain is the input. The identity is the output.
https://t.co/n4vcxHGt9q
Disappointing update, with one silver lining: .twin isn't proceeding with its ICANN application after all — the partner behind it chose to focus on Web3 only instead. Registered these on the strength of the original announcement, so ICANN just flipped to ICANT on this one.
All the anticipation now shifts to .hub, which is still moving forward on the ICANN track.
With all the discussion around the new ICANN applications, two I’m especially interested in are .twin and .hub.
I invested early, believing AI Agents + Digital Twins would increasingly converge.
A few I acquired:
AgentAI. twin
AgenticAI. twin
AgentTwin. twin
Commerce. hub
Fintech. hub
Still speculative, but that’s what makes this ICANN round so interesting.
@spaceship Wanting to make you aware of the marketplace showing an error "unable to apply these settings to an issue on our end" when attempting to update pricing. Thx.
No Prompt. No Templates. Just a Name.
DARWA™ doesn't build a site to fill in. It reads the name and renders the brand — complete, one-of-one, live in seconds.
That's https://t.co/HYY13Xrba4 — start to finish, nothing typed but the name.
The ultra-premium names sell for six figures to a million-plus — on brand authority alone.
But your other 5, 10, 100, 1,000+ names? Dead link, a 2019 "for sale" banner, same "Make Offer" template as every parked domain.
That's who DARWA™ is for.
With all the discussion around the new ICANN applications, two I’m especially interested in are .twin and .hub.
I invested early, believing AI Agents + Digital Twins would increasingly converge.
A few I acquired:
AgentAI. twin
AgenticAI. twin
AgentTwin. twin
Commerce. hub
Fintech. hub
Still speculative, but that’s what makes this ICANN round so interesting.
The inverse has been an interesting design constraint for me. Instead of giving a model more bits, I built a pipeline that starts with the fewest possible bits — a domain name, no prompt — and has to resolve a fully specified brand identity from it. Different problem, same underlying question: How much can be inferred from how little?
Four Frontier AI Labs in three weeks:
·OpenAI
·Anthropic
·Meta
·Moonshot
Different failures, same lesson:
Containment isn't identity.
At Agent scale, "Can it get out?" isn't the only question. The ones that matter:
Who is this? What can it do? Can I trust it?
The internet has no universal answer for that yet. It's going to need one — and it won't be a bigger cage. It'll be an Identity Layer.
David, this is one of the more honest takes on algo valuation I've seen — and I think you're right that GoDaddy's number wasn't "stupid," it was structurally blind. But I'd push on where the blindness actually lives.
The team .ai miss isn't that it used comps. It's that it comped the string "team" across every TLD, saw everything but team .com under $8K, and regressed toward that. The name never gets read for what it is — a short, one-word dictionary term — it just gets matched to its own weak sales history.
Flip the anchor and the error mostly goes away. Classify the name first (short dictionary word = top tier), then price it within the .ai pool — against how other strong one-word .ai names have actually cleared — instead of against "team.*" history. Do that and team .ai lands in six figures, not $5K, because you're pricing the class on the TLD, not the word's past.
Where I'll fully agree with you: even that doesn't front-run a category. Class-first pricing stops you from undervaluing a strong name on a hot TLD — but it still takes a human noticing the TLD itself has re-rated to move the whole framework. No algo I've seen truly predicts the turn; the honest ones just admit which part is data and which part is judgment.
(It's basically why I ended up building my own engine around classification + TLD-class comps rather than name-string matching — but that's a longer post.)
@MediaOptions Owning it and activating it are still two different steps though — most brands sit on the domain equivalent of a locked garage. The name is yours, but it's not doing anything yet. That gap between owned and activated is the interesting part right now.
Yesterday I posted: type a name, watch a brand exist. Today, watch. Unedited. Real time. One take.
No prompt. No templates. No design decisions.
Just a name.
Built solo in South Florida, for the holders who knew a domain was worth more than a parked page. Domains were the original digital real estate — and they deserve better. Build with Fire. Deploy with Purpose.
DARWA™ is live. Type a name. Watch a brand exist — in seconds. No prompt. No templates. No code. No design decisions. The domain is the input. The identity is the output.
https://t.co/n4vcxHGt9q