How to automate list building with AI
A sheet of every company in a directory you chose, each row carrying its source URL, verified as still trading, with the owner named and their published email.
A bought list was assembled against generic firmographics, a chunk of the contacts have changed jobs, and your competitors bought the same rows. You find out after the first send, when bounces run at 14%.
Building your own used to cost hours of directory-crawling per hundred rows. A companion in Strawberry crawls the source completely, verifies each row against a second source, and builds the sheet in your own Google Sheets. The prompts below are the whole method.
Pick sources instead of filters
“SaaS companies, 50–200 employees” is a filter anyone can apply.
“Every company that exhibited at this conference” is a source, and it carries intent no filter can express.
Prompt: “I sell inventory software to independent hardware retailers in the UK. Before we build anything, help me list ten places where these companies are enumerable: trade association member directories, buying-group member lists, conference exhibitor lists, trade magazine award shortlists, franchise directories. For each, tell me roughly how many companies it holds and whether it’s public.”
Do this once per market and you reuse the answer for a year.
Why crawl one source properly first?
Because coverage you can see is worth more than volume you cannot.
The companion goes through every member, including the ones it cannot fully complete.
Prompt: “Start with the BHF member directory. Go through every member and build a sheet: company name, website, town, number of branches if stated, and the directory URL for the row. Don’t skip entries you can’t fully complete. Include them with blanks so I can see the coverage.”
Why verify before you enrich?
A list of dead companies enriched with contact details is an efficient way to waste a week.
Prompt: “For each row, open the company website and confirm it’s trading: recent news, current opening hours, an updated copyright year, live socials. Add a column: ‘active’, ‘uncertain’, or ‘likely closed’, with the reason. Then drop nothing, just sort by that column.”
When should you start finding people?
Once the active rows are known.
Then the companion names the owner or operations manager and finds their published email.
Prompt: “For the ‘active’ rows, find the owner or operations manager: name, title, LinkedIn URL, and their work email if it’s published on the site or in the directory. Where email isn’t published, write ‘not published’. Don’t guess a pattern. Add a confidence column for whether this is the right person to talk to about inventory systems.”
The no-guessing rule costs coverage and buys deliverability. Pattern-guessed addresses are how sender reputation dies.
Keep it fresh
A monthly routine turns the list into an asset instead of a snapshot.
Prompt: “Every month, re-check the directory for new members and add them in the same format. Also re-verify any row I’ve marked ‘contacted’, if the contact changed jobs, flag it. Send me the additions and the changes only.”
It is the same prospecting loop our sales users run, pointed at whatever market you sell into.
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