Automated candidate sourcing from every place your next hire already is
Automated candidate sourcing means software finds people who fit a role for you. RecruitingBot turns your rubric into searches across public professional profiles, open web sources, your own talent database and an apply page, then removes duplicates and logs why each person is there.
- GDPR built in
- Human approves every shortlist
- No hiring fees
Four pools that feed one candidate list
Most sourcing stalls because each pool lives in a different tab. Your past applicants sit in the ATS, silver medalists from last year sit in a spreadsheet, fresh inbound applications land in a shared inbox, and the passive market lives on public profiles you search by hand. RecruitingBot pulls all four into one list per role, so the same person is never scored twice and nobody falls between systems.
Each pool has its own rules. People who applied gave you their details for this purpose. People in your own database were collected under your existing policy. People found on public profiles or the open web never contacted you, which is why they are handled under documented legitimate interest and receive a notice in the very first message. The source column on every candidate tells you which of those rules applies.
| Source | What it adds | How it gets in |
|---|---|---|
| Public professional profiles | Passive people who match the title, skills and location | Searches built from your rubric, run by the sourcing step |
| Open web sources | Portfolios, code repositories, talks, published work | Matched to the same person where the signals agree |
| Your talent database | Past applicants, referrals, silver medalists | CSV upload on every plan, ATS import from Greenhouse, Lever or Workable on Growth and Agency |
| Hosted apply page | Active job seekers who found the post | Each role gets a public page with a short form and spam protection |
Search strategy written from your rubric
A good sourcer does not type the job title into a search box and call it done. They translate the brief into synonyms, adjacent titles, must have skills, likely previous employers by type, and exclusions. RecruitingBot does the same translation from the role intake. A "Senior Data Engineer" brief with must haves for streaming pipelines and a warehouse becomes several query variants: analytics engineer with streaming experience, platform engineer who owned ingestion, backend engineer with heavy data work.
Deal breakers become filters rather than afterthoughts. If the role is onsite in Austin and relocation is not offered, people far from Austin who never mention relocating are dropped before they cost a contact reveal. Nice to haves widen the net slightly without changing the order of priorities. You can open the generated search plan, remove a variant that is too broad, add a title your industry uses, and rerun only that branch.
- 01 Title variants. Adjacent and older titles for the same work, so people with unusual job names are not missed.
- 02 Skill clusters. Groups of tools and methods that usually appear together, which catch people who list only some of them.
- 03 Seniority signals. Years in role, scope of ownership and team size, used to keep juniors out of a lead search and the reverse.
- 04 Exclusions. Current employer of the client, people already in an active process and anyone on your suppression list.
Why the same person shows up only once
Duplicates waste money in two ways. You pay to reveal the same contact twice, and worse, a candidate gets two different emails about one role from the same company. RecruitingBot matches people across sources using name, location, current role, profile links and email where available. When a public profile and a CSV row point to the same person, they merge into one record that keeps both source entries, so the history stays honest.
Dedupe also runs across roles in a workspace. If someone is already in the pipeline for your backend role, they appear with a flag on the data role instead of a fresh outreach, and you decide whether to move them. For agencies running several client workspaces, matching stays inside each workspace on purpose, because one client's pipeline is not something another client should see.
Lawful basis logged before anyone is contacted
Sourcing passive candidates in Europe is lawful when you can show a legitimate interest, keep the processing proportionate, tell people early and let them leave easily. That last part is where manual sourcing usually breaks down. RecruitingBot records the source and the lawful basis on every candidate at the moment they enter the list, and each role gets a legitimate interest assessment template you can fill and keep.
The first message to a sourced person includes an Article 14 notice covering who you are, why you are writing, where the details came from, how long you keep them and what rights they have. Every message carries a one click opt out that writes to a workspace wide suppression list. Sourced people who never reply are deleted after 30 days by default, adjustable between 30 and 180. More detail lives on the GDPR compliant candidate sourcing page.
- Lawful basis
- Legitimate interest, assessment v2 attached
- Source
- Public professional profile
- Article 14 notice
- Included in first message
- Opt out
- One click, suppression list
- Retention
- Deletes in 30 days without reply
Contact reveals are a budget, not an afterthought
Finding a profile is cheap. Finding a working email address for it is the expensive part, which is why every plan has a monthly pool of contact reveals: 300 on Starter, 800 on Growth and 2,000 on Agency. RecruitingBot spends that pool only on people who already passed your deal breakers and scored well enough on the rubric to be worth writing to. Nobody gets revealed just because they appeared in a search.
Imported candidates and applicants usually bring their own email, so they do not touch the reveal pool at all. You see the pool as a meter in the dashboard and a prompt when you reach 80 percent. Reveals count consumption, so removing a candidate after revealing them does not give the reveal back. If you regularly run out mid month, the AI recruiter pricing page shows the higher limits and the yearly price.
| Plan | Contact reveals per month | Candidates analyzed per month | Active roles at once |
|---|---|---|---|
| Starter | 300 | 1,500 | 3 |
| Growth | 800 | 5,000 | 8 |
| Agency | 2,000 | 12,000 | 20 |
What a sourcing run looks like on a real role
Picture a startup hiring an account executive for mid market deals, remote within US time zones, with a salary band and a must have of closing annual contracts above a set size. In a typical example run, the sourcing step finds around 500 people across public profiles, the company's CSV of past applicants and the apply page. Rubric filters narrow that to about 80 selected, and only those 80 are considered for reveals and outreach.
From there the rest of the pipeline takes over: personalized emails from the founder's connected mailbox, reply handling, screening questions and a short skills test. In that same example, 25 people reply and 5 reach the shortlist. Those numbers are an illustration of the funnel shape, not a promise, because every role, market and salary band behaves differently. See the full flow on how AI recruiting works.
-
Day 1
First messageSent
Personal note on their payments work, role facts, salary range, opt out link
-
Day 4
Follow upSent
Short nudge with one new detail about the team
-
Day 9
Last noteScheduled
Polite close, door left open
When automated sourcing is the wrong tool
Automation shines on knowledge work roles where people describe their work publicly: engineering, data, product, design, sales, marketing, customer success, operations, finance, HR and support. It is a weak fit for roles where the talent pool barely exists online, for example certain licensed trades, hourly shift work or very senior executive searches that run on private relationships. For those, a referral network or a specialist search firm usually wins.
It also does not replace judgment about the brief itself. If the salary band is far below market or the must haves describe three jobs at once, a faster search only surfaces that problem sooner. RecruitingBot shows you how many people pass each filter, which often tells you more about a brief than a week of unanswered messages would. Loosen one requirement, rerun, and compare the counts before spending reveals.
Questions buyers ask
Anything else goes to [email protected] or the help panel inside your dashboard.
Is an AI sourcing tool the same as a people database?
No. A people database gives you search filters over records it already holds, and you do the rest. RecruitingBot builds the searches from your rubric, combines public results with your own database and inbound applicants, removes duplicates, scores everyone and then hands the strongest people to outreach and screening inside the same pipeline.
Can I source only from my own database and applicants?
Yes. You can switch off public sourcing for a role and work only with imported candidates and people who apply through the hosted page. Many agencies do this for rediscovery, scoring last year's applicants against a new brief before looking outside. Imported records rarely need contact reveals, so the reveal pool stays untouched.
Which sites does RecruitingBot search?
It works with public professional profiles and open web sources such as portfolios, published work and code repositories, under documented legitimate interest. We do not publish a list of individual sites. What matters for you is that each candidate record shows its source and lawful basis, which you need when someone asks where their details came from.
What happens to sourced people who never answer?
They are deleted automatically after 30 days by default. You can set the retention period anywhere from 30 to 180 days per workspace. Anyone who opts out goes onto the suppression list immediately, so a later search will not bring them back into a pipeline or send them another message.
Do duplicates use up my monthly limits?
Duplicates are merged before scoring, so the same person found in two places counts once toward candidates analyzed and never triggers a second contact reveal. Limits count what is actually used during the billing period, and deleting a candidate later does not return analyses or reveals already spent.
Can I import candidates from my ATS?
CSV upload works on every plan. Growth and Agency also import from Greenhouse, Lever and Workable, and Agency adds API access. Imported people are matched against existing records, scored against the role rubric and can go straight into outreach without spending reveals when their email is already on file.
Keep reading
Give it a role and see the shortlist it builds
Type a job title on the homepage. The demo runs instantly without an account and shows an example pipeline, job post and shortlist for that role.