The problem this solves
Every founder can name their best customers instantly. Turning that intuition into a prospect list is where the process breaks. 'Companies like Acme' is a pattern in someone's head: a size band, a stack, a motion, a stage. Extracting the pattern, expressing it as searchable criteria, and then actually searching is analyst work that most sales teams never get around to, so they default to broad filters that produce lists nothing like the customers they love.
Even when someone builds the lookalike list, it usually arrives as bare company names. A name without reasoning forces every rep to redo the qualification: why is this company here, what does it have in common with our best accounts, what is the angle? Multiply that by fifty rows and the list sits unworked. The match explanation is not decoration; it is the difference between a list that gets called and a list that gets archived.
And the list has to live where the deals live. For a Pipedrive team, that means organizations created cleanly with notes attached, so the first touch starts from context instead of from a blank record. A lookalike program that ends in a spreadsheet asks every rep to do the import themselves, and the quiet truth of sales tooling is that nobody ever does.
How the mission runs
- Profile the seed customers. The agent studies your three named customers - their size, industry, funding history, and business model - using Firmographics and Web Research to build an explicit profile of what makes them alike. The pattern in your head becomes written criteria you can inspect and correct before any searching starts.
- Read their stacks with Technographics. Technographics captures what the seed customers actually run, because shared tooling is often the strongest lookalike dimension: it encodes budget, sophistication, and adjacent needs simultaneously. The stack fingerprint joins the firmographic profile as a second axis of similarity, giving the search two independent dimensions to match on.
- Search for the resemblance. With the profile fixed, the agent searches for companies matching it across both axes, scoring each candidate on how many seed traits it shares. The 50 strongest matches survive, drawn from beyond the obvious keyword-adjacent names a manual search would have found.
- Enrich every match. Each of the 50 is enriched with current employee count, funding stage and latest round, and detected technology stack, so every record arrives with the shape a rep needs for a first-touch angle already visible. Enrichment happens at build time, so the numbers reflect the company as it is today.
- Create organizations in Pipedrive with the why. The agent writes each company into Pipedrive as an organization, with a note explaining the match: which seed customer it resembles and on which traits. The batch is staged for your approval before writing, and existing organizations are detected so nothing duplicates.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
Fifty organizations in Pipedrive, each enriched with size, funding, and tech stack, and each carrying a plain-language note naming which of your best customers it resembles and why. The seed profile itself is included in the mission summary, so you can see the pattern the agent extracted and tighten it for the next run. Reps open each record to a ready-made qualification instead of a bare name.
Make it yours
- Seed with different customers per segment - your three best enterprise logos and your three best startups - to build two distinct lookalike pools.
- Raise the count to 150 and add a scoring column instead of a hard cutoff, letting your team choose their own depth.
- Deliver the batch to a spreadsheet for a manual scoring pass first, then promote only the top tier into Pipedrive once the pattern has proven itself.
Frequently asked questions
What makes a company a lookalike here?
Resemblance is scored across the traits extracted from your seed customers: size band, industry, funding stage, business model, and detected stack. A candidate matching four traits outranks one matching two. The extracted profile is shown to you, so the definition of similarity is inspectable rather than a black box.
Will this pull in companies we already work with?
The agent checks candidates against your existing Pipedrive organizations before the write and flags overlaps instead of recreating them. Current customers and open deals are excluded from the new-record batch, and the staging review gives you a final pass regardless.
How do I make the next run better?
Correct the profile. The mission summary states the criteria it derived from your seeds; if a trait is wrong or missing, say so in the next prompt and the search re-runs against the corrected pattern. Two or three iterations typically converge on a list your team fully trusts.