Service

Autonomous lead generation

Buying a list gives you ten thousand rows and a bounce rate. What a sales team actually needs is forty companies this week that genuinely look like the last customer who signed — with a reason attached to each one.

Where the leads come from

From the open web, gathered the way a well-behaved crawler gathers it: company registries, trade directories, association membership lists, tender portals, exhibitor lists, job postings, and the companies’ own sites. We respect robots directives and rate limits — partly because it is the right way to operate, and mostly because aggressive scraping gets your infrastructure blocked within a fortnight and the pipeline dies.

We do not scrape platforms that forbid it in their terms, and we do not sell you data harvested from walled gardens. If someone offers you that, the risk transfers to you, not to them.

The part everyone skips: verification

A raw list is mostly noise. Before anything reaches your CRM, each record is checked: does the company still exist, is the site live, is the address plausible, does the email domain accept mail, is this actually a manufacturer or a reseller pretending to be one. Roughly half of what a first pass collects is discarded here, and that discard is the product.

Scoring against your real customers

We start from the customers you already have and won profitably. From those we derive the signals that actually correlate — company size band, sector, export markets, technology in use, recent hiring, a product page that mentions the exact material you supply. Every lead arrives with a score and, more usefully, a sentence explaining the score.

A lead your team does not trust is worse than no lead. The first thing a sceptical salesperson does with an AI-generated list is find one bad row and dismiss the whole thing. So every score is explainable, and low-confidence rows are labelled as such rather than padded in to make the number look bigger.

First contact, in their language

Outreach is written per company, referencing something real from their site — a market they have just entered, a certification they hold, a product line that matches yours. It is written in the recipient’s language by a model tuned for that language, not translated word-for-word from English into something that reads like a scam.

Volume is capped deliberately. Sending is throttled, suppression lists are honoured, and replies are routed to a human. We build outreach that could be read aloud on a call without embarrassment — which happens to be the same thing as outreach that does not destroy your domain reputation.

What you get each week

A shortlist of verified companies, scored, with the reasoning attached.
Contact routes that have been checked, not guessed from a first-initial pattern.
Drafted first-contact messages in the target language, ready for a human to approve.
A record of what was sent, to whom, and what came back.

What we will not claim

We will not tell you a conversion rate before we have run your market. Anyone who quotes one on a first call is quoting someone else’s. What we can tell you within two weeks is how many qualified companies exist in a given market and how many are reachable — which is the number that decides whether the channel is worth building at all.