← all playbooks

Find more candidates like the best person you ever placed

A written description of what defines your seed person, corrected by you before any search runs, and a ranked list of people who share those traits, each showing which traits they match and which they miss, with dated work history and a LinkedIn profile.

use when
you can point at one person and say more like this, faster than you can write the brief that describes them
starts from
A candidate

The prompt

Paste it into Claude Code or the Claude desktop app with Hyreflow connected. The first line tells your agent to use Hyreflow, so it reads the play, asks before it spends anything, and hands the work back to you.

paste this into Claude
Use hyreflow and find me more people like this one.

The person: <LINKEDIN_PROFILE_URL_OR_NAME_AND_EMPLOYER>. I placed them at
<CLIENT_COMPANY> and they are the best hire I made there.
I want lookalikes in <LOCATION> for <ROLE_OR_DESK>.

How I want it run:
- Pull their real work history first. Then tell me in plain words what you
  think defines them: employers, kind of company, career path, skills.
- Stop there and let me correct it before you search. I will tell you which
  traits matter and which are an accident.
- Never use age, gender, nationality or anything like them as a trait.
- Search on the corrected traits. Leave out the person themselves and
  anyone at <COMPANIES_TO_EXCLUDE>.
- Pull work history for what comes back and score each person trait by
  trait, so I can see why they are on the list.
- Contact details for the closest matches only: personal email and
  LinkedIn, never work email.

I want <N> people. Run five first and tell me what the full run costs
before you do the rest. Do not contact anyone.

Replace every <PLACEHOLDER> with your own detail. Everything else can stay as written.

What you need first

  • One person to use as the template: a LinkedIn profile link, or a name and employer
  • The market you want the lookalikes in: the location, the desk, and any employers to keep out
  • A Hyreflow workspace with credits

Tools it can reach for

The agent picks per step from what your workspace has. Nothing here is required by name.

What happens when you run it

Free steps are marked free. Anything that spends credits is marked, and the agent asks before the first paid run of any size.

  1. 1

    Name the seed person

    free

    You give a LinkedIn profile link, or a name and an employer. The seed is a template only. They are not contacted and they are removed from the results.

  2. 2

    Pull the seed's dated work history

    credits

    If you gave a name, the profile is found first. Then employers, titles and dates are pulled for the seed, plus the headline, skills and role descriptions where the provider carries them. A seed with no dated history gives nothing to work from, and the agent says so.

  3. 3

    Write down what defines them

    credits

    The agent turns the history into a short list of traits: the employers and the kind of company, the path between roles, how long they stay, the words their profile uses for the work, the skills. Protected characteristics are never traits.

  4. 4

    Correct the trait list

    free

    You see the list before any search runs. Strike what is an accident of where they worked, add what you know and the profile does not show, and mark the traits that are not negotiable.

  5. 5

    Search the databases on the traits

    credits

    A structured people search built from the corrected traits: the employers the seed came through and companies like them, the titles on that career path, the location. The count is read before the pull.

  6. 6

    Search on meaning as well

    credits

    The same traits written as a plain description and run as a meaning-based search for people, which reaches profiles that use other words for the work. Each search is charged whether or not it finds anyone, so it starts with a few results.

  7. 7

    Pull work history and score trait by trait

    credits

    Both result sets merge on the LinkedIn profile. Dated history is pulled for the pool, then each person is scored against the corrected traits one at a time, so the list shows why someone is on it and what they lack.

  8. 8

    Find contact details for the closest matches

    credits

    Personal email and LinkedIn for the top of the list only, through the personal-email providers in a set order that stops at the first hit. A miss usually costs nothing.

What a run costs

Credits are spent per candidate the play actually works, and a lookup that finds nothing usually costs nothing. The two figures are the run where the first provider answers and the run where every lookup walks its full chain.

candidatesif the first provider answersif every lookup walks the chain
25$3.434 credits$13130 credits
100$14140 credits$53530 credits
500$68680 credits$2632630 credits
1,000$1351350 credits$5255250 credits

Free before anything is charged

  • Name the seed person
  • Correct the trait list

What moves the number

  • The channel. This play buys personal email, because these are candidates, and a work inbox is the wrong place to approach one. A personal address costs more to find than a work one.
  • Coverage on work history, people search and personal email. The chain stops at the first provider that answers, and only that provider bills.
  • How many candidates survive the free filters. Everything dropped before the paid steps costs nothing.
  • The scoring and drafting steps run on the metered agent, charged on what they read and write rather than per candidate, so they sit outside this table.
  • Providers you connect with your own key. Those calls bill your account, not your credits.

An estimate, not a quote, priced at the volume credit rate. Your agent sizes the run against your own workspace and tells you what it will cost before it spends anything.

Pointing at the right person is faster than describing them

Every recruiter has a placement they would clone. Ask them to write the brief for that person and you get a job title and three adjectives. Ask them to point and they are certain. "More like this one" is the most natural sourcing instruction there is and the hardest to hand to a database, because a database takes filters and a person is not a filter.

This play does the translation in the open. It pulls the seed person's dated history and writes down what appears to define them: where they worked and what kind of company that was, how their career moved, how long they stay, what their profile says about the work. Then it stops and shows you. That is the step that matters. Some of what sits on a profile is the reason they were good, and some is an accident of where they happened to work. Only you know which is which.

There is no find-similar button under this. No provider here takes one person's profile and returns similar people, and the similarity lookup that exists works on web pages and suits company sites. So the agent searches on the corrected traits, once in the databases and once on meaning. You can see what similar means, and you can change it.

What you get back

The trait list as you corrected it, and a ranked CSV. Each row has the person, their current title and employer, the traits they match and the ones they miss, the dated history behind that, which search found them, the LinkedIn profile, and a personal email for the closest matches.

The seed is not in the list and nobody has been contacted.

Variations worth knowing

Several seeds. Give three strong placements and the traits they share carry more weight than the traits of any one of them.

One angle among five. Map a whole talent pool runs lookalike expansion beside four other angles and picks its own seeds from the best people it finds. This play is that angle by itself, with a seed you chose.

Keep clients and live processes out. For more than a couple of exclusions, run the result through your off-limits rules before anyone is approached.

Where this goes wrong

What made them good is not on the profile. A search can only copy what a career shows. The rest goes into the call questions.

Copying the accident. Left uncorrected, the search clones the seed's employer list and misses the point. The correction step is free, so use it.

A thin seed. A profile with no dated history yields no traits. The agent asks for a second seed or a written brief and does not guess.

A narrower pool on purpose. Lookalikes share a background. Treat the list as one angle on the market and keep a normal search running beside it.

Questions

Is there a find-similar button underneath this?

No. Nothing in Hyreflow takes one person's profile and returns similar people. The similarity lookup that does exist works on web pages and is used for company websites. So the agent derives the traits and searches on them. The upside is that you can read what similar means and change it, which a black box never lets you do.

What if the thing that made them good is not on their profile?

Then a search cannot find it. Drive, judgement and how someone handles a difficult client leave no trace in a work history. The correction step is where you add what you know, as long as it can be tested against a career: always stayed through a full product cycle, say, or only ever worked in founder-led firms. Anything else becomes a question for your first call.

The seed is someone I placed. Is that a problem?

Not as a template. The seed is never contacted and is removed from the results. Their current employer is your client, so name it as a company to keep out and nobody is sourced from the firm you placed them with. Exclusions are applied to the results at no cost.

Will this just hand me clones with the same background?

It narrows by design, because that is what a lookalike is. Traits are limited to skills, experience, the scope of past roles, employers and career path. Age, gender, nationality and anything like them are never traits and never scored, whatever the seed looks like. Run it beside a normal search, not as a replacement for one.

Does it work better with more than one seed?

Usually. A trait shared by three of your strong placements is a signal, and a trait only one of them has is probably an accident. With five or more seeds a provider-side people lookalike search becomes an option as well. It needs at least five seed profiles and bills per person returned, so it suits a desk with a track record, not a single example.