Our own sales engine.

Searches the map for a trade in a city

Our own copy of a map scraper collects businesses with their address, phone, site and reviews. Then it hunts for a contact email for each one.

• Looks on the business’s own site, in web archives, and on its Facebook and Instagram pages
• Filters out addresses that are not a person or a business inbox
• Checks that the address’s domain can receive mail before keeping it

Scores each business against its local competitors

Four areas are scored: visibility, reputation, website and conversion. Each business is compared with the businesses near it, not with a generic benchmark.

• A check that cannot run is reported as unverified, never guessed
• Only what was measured goes into the report
• Runs around the clock in the background

A person releases every batch

An audited business is only emailed after the operator approves its batch. A business whose audit is being refreshed is held, not sent a stale message.

Runs an email platform that has no API

The platform offers no way to be driven by software, so an unattended robot drives it through a browser: it uploads leads, pauses and resumes the campaign, and reports back.

• Each lead gets a first message and two follow-ups, with follow-ups sent first
• Daily caps per sending domain keep each domain’s reputation safe
• The operations cockpit shows no number it cannot trace to a source

A knowledge base every AI session reads first

One note per system (30 notes), a map of how the business’s things relate, and a log of decisions, kept as plain files in the project. Work on a system starts with its note, so nobody rebuilds the model from the code.

One short list each morning

Reads the status of every machine and writes what needs the owner today, and what it already handled on its own.