Lookalike engine
One domain in.
Ten thousand like it out.
Find companies similar to your best clients. Paste one domain — a customer you'd happily clone — and get thousands more that sell the same thing to the same kind of buyer, closest match first. No filters to guess at, no industry codes to argue with. And you see your matches before you give us anything — the search below works right now.
Real rows from the live index · no email, no card, nothing to install
Companies indexed
60M+
Rows per search
10,000
Flat price
$49/mo
Searches
Unlimited
How it works
[ 01 ]
Paste a domain
Your best customer, your loudest competitor, the account you wish you had ten more of. Nothing else: no list to upload, no filters to guess at, no account.
[ 02 ]
We embed it
The company's description runs through the same embedding model the whole database went through. A domain we've never seen gets read off its own website and embedded on the spot — that's the few seconds you wait on the first search.
[ 03 ]
The graph ranks itself
One nearest-neighbour query over 60M+ company vectors comes back ordered, with a similarity score on every row, so you can see exactly where the resemblance runs out.
How it's built
Every company in the graph carries a description — what it sells, to whom, in its own words. Those descriptions are embedded with OpenAI's text-embedding-3-large at 1,024 dimensions and stored in Turbopuffer, one vector per domain. A search embeds your seed the same way and asks for its nearest neighbours by cosine similarity, which is why the results are ranked by resemblance rather than by whichever tag a data vendor happened to attach. Matches below 0.65 are dropped. Seed a domain we have not indexed and we scrape its site and embed it on the fly, then throw the query at the index — same path, a few seconds slower.
Questions
How is this different from an industry filter?
An industry filter matches a label somebody else assigned. This matches meaning: the seed company's own description of itself against every other company's, in the same vector space. It finds the competitor filed under the wrong SIC code, the company that describes the same job in different words, and the adjacent business your ICP spreadsheet never had a row for.
Where does the company data come from?
The same graph behind the rest of the site: 60M+ company records assembled from bought datasets, public filings and our own scrapers, with descriptions, industry, headcount band, country and website on each one. The vector index is built from those descriptions.
Can I filter the lookalikes by country or size?
Yes. The engine takes country and company-size filters and applies them inside the vector query, so you get the closest 10,000 companies that also fit the constraint rather than a shortlist filtered down to nothing. The free preview runs unfiltered.
How do I get the whole list, with contacts?
A subscription: $49 a month, unlimited searches. Every seed you paste returns up to 10,000 ranked companies with the decision-maker and their email on each row, and the whole thing exports to CSV. No credits, no per-record pricing, cancel anytime. It is also included in Pro ($199/month), which adds every other dataset plus the API and MCP.
How accurate is the similarity score?
It is cosine similarity between two embeddings, nothing more and nothing dressed up. Above roughly 0.80 the matches usually read as obvious competitors; 0.70 to 0.80 is the adjacent-market band that is often the interesting part; the engine stops at 0.65 rather than padding the list with noise. Judge it by the rows, which is why they are on this page before any signup.
What if my domain isn't in the index?
Then we fetch and embed it live, which is why a cold domain takes a few seconds instead of being instant. If the site gives us nothing to read — a parked domain, a login wall, a one-line splash page — the search comes back empty and a better-known domain is the fix.
The offer
$49/month
- Unlimited searches — every seed returns up to 10,000 ranked companies, no credits, no per-record pricing
- The decision-maker and their email on every row, plus unlimited CSV export
- Cancel anytime — everything you exported stays yours
No signup to look · sign up when you want the whole list
Already on Pro ($199/month)? Unlimited lookalike search is included — along with every other dataset, the API and MCP.