What it does
Leafer reads public posts on Google Maps, Reddit, X, LinkedIn, Facebook and Instagram and looks for one thing: a person who has already described, in their own words, the problem a product solves. Someone complaining about a tool is a better prospect than someone who matches a firmographic filter, because they have already done the hard part — deciding something is wrong.
The three steps
Find searches the six channels in parallel, classifies the signals it gets back, scores them and removes duplicates. The same complaint posted to two platforms is one person, not two prospects.
Prove crawls the company site behind a signal, works out who makes the decision, and verifies a work email. A signal without a reachable person is a screenshot, not a lead.
Draft writes the first message in the product's own voice and puts it in an approval queue.
What is now being built
Sending. Discovery, enrichment, scoring and drafting run today; outbound delivery is the piece still under construction, and the approval queue is the shape it is being built around rather than a gate to be removed once it works.
A tool that finds someone mid-complaint and answers automatically produces the exact reply everyone has already learned to ignore. So the queue stays: a person reads the draft and decides whether the post deserves an answer at all. That is the slower version and the one that survives contact with the people on the other end.
It is being built last for the same reason it is being built carefully. It is the only part of this that arrives in someone else's inbox, and the only part that cannot be taken back once it goes out wrong.
Where the data sits
In the EU, and only what a public post already contains.