Find similar companies from your best customers
When a company becomes a customer, signs up, or visits your site, the workflow finds similar companies and removes the ones your team already knows.
The situation
Every good customer teaches sales something: this type of company buys. The same is true when a promising company signs up or visits the site. Usually, that evidence goes nowhere. A deal closes, the team celebrates, and the next prospect list comes from a broad filter unrelated to what worked.
The useful question is simple: which other companies look like this one? Finding the answer by hand means repeating the same research every time a useful signal appears.
The constraint
Company descriptions are not always good search inputs. They are marketing copy: full of positioning, category language, and claims about outcomes. A similarity search based on that raw text finds companies that write in a similar style, not necessarily companies that do similar work.
The workflow first turns the raw LinkedIn description into a plain search profile of what the business actually does. The model shapes the search input; it does not decide which companies are good targets.
Two matching methods run together because they find different kinds of companies. Domain matching identifies businesses with a similar technical or commercial shape. Semantic matching finds businesses doing a similar job under different language.
The final constraint is duplicate work. Every candidate is checked against Airtable and HubSpot before enrichment. A company already discovered or already in the CRM is not researched or paid for twice.
What we built
A trigger arrives with a company domain, LinkedIn description, the source signal — website visitor, signup, or closed-won customer — and the Slack thread that requested the work.
The description becomes a simple search profile. Domain matching and semantic matching find potential lookalikes. Results are processed one at a time, invalid responses are discarded, and the rest are normalised into consistent fields.
Every candidate is checked against Airtable and HubSpot before enrichment. New companies receive LinkedIn enrichment, land in the shared sales sheet, and are returned to the same Slack thread where the request started.
Process diagram
Where it stands
Closed-won customers, signups, and qualifying site visits can become starting points for finding new accounts. Sales asks in Slack and receives enriched, deduplicated companies in the same thread.
Next
Failed payment recovery that catches repeat failures
Ready to boost business results with AI automation?
Book a 30-minute operations review. We will look at where manual work is limiting your operation and whether there is a practical opportunity to create capacity.
Book a 30-minute operations reviewSwiss precision. Valley speed.