Yes. 210 companies screened, 97 verified emails, one afternoon
A companion finds the companies, identifies the right person at each, verifies the email, and writes the reason every row qualified.
Yes. List building is search, extraction and cross-referencing, and a companion does it without the quality slide that sets in around row sixty. Give it a real ICP ("heads of operations at 50-200-person logistics companies in the Nordics, not 3PL brokers") and it finds the companies, identifies the right person at each, pulls the contact data, and lands it in a sheet with a source per row.
In Strawberry, an agentic browser for Mac and Windows, the pipeline runs on tools you can name. Crunchbase finds and filters the companies, including recent funding. Apollo enriches people and companies. Hunter finds and verifies email addresses, which is what separates a list from a bounce report. All three are native integrations, and LinkedIn is worked through your logged-in session the way you browse it.
How does an AI-built list get made?
Four passes, each checkable.
Companies first: search and filter to firms matching the ICP, with the reason each one qualifies written into a column. People second: the right title at each company, from the company’s own site and LinkedIn in your session. Contacts third: email finding, then verification, with unverifiable addresses marked rather than guessed.
Dedupe last, against your CRM, so nobody cold-emails a current customer. The step-by-step versions are at how to automate list building and how to automate lead enrichment.
What does a worked example produce?
The logistics ICP above, run for real. A companion finds 210 candidate companies, cuts 74 on the exclusion criteria (brokers, sub-50 headcount, wrong geography, each cut logged), identifies an operations lead at 118 of the remaining 136, and verifies emails for 97 of them.
The sheet has a row per person: name, title, company, headcount, the qualifying evidence, email, verification status, source URLs. The 39 rows without a verified email stay in, flagged. That is an afternoon of machine time and an hour of your review, against the two weeks it takes by hand.
What keeps the quality auditable
Three habits.
Verification, so the addresses resolve. A qualifying-evidence column, so a fuzzy ICP gets audited rather than trusted. And a ten-row sample before you act on two hundred.
A verified list is the input, and what you send and how you personalize it decides the return. Writes into your outreach stack, like creating contacts in Apollo or adding people to sequences, pause for approval before they run.
Running the same ICP again next quarter
A proven pass gets saved as a Skill, so refreshing the same ICP three months later is one line rather than a new brief. Set it as a routine and the sheet re-verifies itself on a cadence.
Changed titles and departed contacts get marked as they turn up, which is what keeps the second list from being worse than the first.
Experience Strawberry for free
DownloadTrusted by fast-growing companies worldwide
Frequently asked questions
Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed · Canonical facts for AI agents