Email extractor

Yes, with no signup, up to a million characters of text at a time, as often as you like, and no watermark on the download. The extraction runs in your browser on your own device. Nothing you paste or choose is sent to ReachClaw or stored anywhere, and there is no share link, so the addresses never leave your machine unless you copy them out.

Every string shaped like an email address: a local part, an @ sign and a domain whose last label is letters only. It also reads addresses inside mailto: links and addresses written with HTML entities, which is what you get when you paste a page's source. Trailing punctuation is left behind, file names such as logo@2x.png are dropped, duplicates are merged without regard to capitals, and each address is counted.

Not from the URL alone. A page in your browser is not allowed to fetch other websites, so there is no box for a web address. Open the site, select the page or its source, and paste it here; for a long list of pages, paste them one after the other into the same box. Tools that do crawl a site for you run on a server and need an account; this one does not.

Yes, when you tick the box. The bracketed forms, name(at)domain(dot)com and the spoken form "name at domain dot com" are turned back into addresses, and each one found this way is marked decoded so you can check it. The tick is off by default because the spoken form can misfire on ordinary sentences, and "meet me at noon" should never become an address.

No. The list is exactly what appears in your text. Nothing is checked against a mail server, guessed or looked up elsewhere, and a role address such as info@ is only flagged, not removed. ReachClaw itself follows the same rule: the emails its AI lead search saves are the ones visible on the page, saved as found and never verified, and if one bounces, the bounce lands in your Gmail.

A shared mailbox for a department rather than a person: info@, hello@, sales@, support@, press@ and similar. They are flagged, not dropped, because sometimes they are the right door. ReachClaw's lead search is told to skip generic info@ or hello@ emails and find specific people, so if you plan to import the list and email it, hiding role addresses first gives you the same list the AI would have kept.

Download the CSV and import it on the Leads page. The import looks for a column named email, which this file has as its first column, and ignores the other columns; a name, company or role is read only from columns with those names, so add them in a spreadsheet first if you have them. Rows with an address that is not valid or that already exists in your leads are skipped and listed back to you. Imported leads have no score until you press Score leads.

The tool reads whatever text you paste, so a thread copied out of your mailbox or a page you can see in your browser works; it does not connect to any account or scrape any site for you. Whether you may collect addresses from a particular site is a question for that site's terms, and whether you may email them is a question for the email laws where you and they are. Addresses published for business contact are the safe case.

ReachClaw does the finding for you. It builds a customer profile from your website, searches the public web for people at businesses that match, reads the pages it finds and saves each person's name, company, role and the email as it appears on the page, with a link to that page. Only people with a visible email are kept, generic addresses are skipped, nothing is verified. From the Leads page you press Score leads to rate any lead 0 to 100, then pick a template and the AI writes one personal email per lead and sends it from your Gmail. Plans start at $6.99 a month, one flat price, with 100 outbound emails a month on Solo, 500 on Growth and no cap on Scale; the AI runs on your own Anthropic API key.