By Andrew Schuessler, Co-Founder of Recruitcha GTM

If you work in recruiting and have not played with Juicebox yet, go try it. Seriously. Type a plain English sentence describing the candidate you need, and watch it return ranked profiles from a pool of hundreds of millions of people. No Boolean strings, no operator gymnastics, no praying that your NOT clause did not just exclude your best candidate.

Juicebox, originally launched as PeopleGPT, proved something important to our entire industry: AI-native sourcing works. Natural language beats Boolean. Aggregating profile data from dozens of sources beats living inside one platform's walled garden. The recruiters who adopted it early got a real edge on everyone still hand-crafting search strings.

So this is not a takedown. It is the opposite. It is about what happens when you fall in love with that way of sourcing and then point it at the searches that actually pay the bills: the hard-to-find candidates, at real volume. Because that is where we hit the wall, and that wall is why we built what we built.

Hard-to-Find Is a Volume Problem in Disguise

First, a truth about difficult searches that took me years on a desk to fully appreciate.

When the role is a controls engineer with specific PLC experience in a fifty mile radius of a small manufacturing town, or a plant manager who has lived through a specific type of turnaround, the talent pool is tiny and almost entirely passive. These people are not applying to anything. They are not on the job boards. Most have not touched their profile in three years.

Filling those searches is not about finding the one perfect profile. It is about achieving total coverage of a small market: identifying every plausible person, enriching every record until you have a real way to reach them, and contacting all of them, professionally and persistently, until the two or three genuinely open ones surface. Hard-to-find recruiting is a coverage and outreach problem wearing a search problem's clothes.

Which means the tool you use cannot just be smart. It has to be smart at volume.

The Credit Meter Problem

Here is the experience every serious sourcer eventually has with AI search tools.

The searching feels limitless. You can run queries all day, refine, iterate, explore the market. The magic is real. Then you go to actually work the market, pull contact information, export the full list, push records into your outreach workflow, and you meet the meter.

Per-seat plans come with allotments of contact credits and export credits. Caps on active projects. Limits on connected mailboxes. The generous tiers cost more per seat, and the truly unlimited tiers live behind custom enterprise pricing. None of this is a scam. It is the SaaS model doing what the SaaS model does: the data is the product, so the data is metered.

For a recruiter working a few searches with healthy candidate pools, the meter rarely bites. But run the coverage math on hard-to-find roles across a full desk, or across an entire agency's open searches, and you are touching thousands of enriched, validated candidate records every month, continuously. At that volume the meter stops being a pricing detail and becomes the bottleneck. You find yourself rationing contact pulls, deciding which qualified candidates you can afford to reach this month. Rationing coverage on a search where coverage is the entire strategy.

What the Power Actually Is

When we sat down and asked what we loved about the Juicebox way of sourcing, it broke down into four capabilities.

Natural language understanding of who you are looking for. Aggregation across many data sources instead of one platform. AI that reads profiles and companies the way a strong researcher would. And speed from intake call to longlist.

Here is what we realized: every one of those capabilities is now available as infrastructure. The same advances that made Juicebox possible make it possible to assemble those capabilities yourself, tuned to your niche, without the meter.

So that is what we did.

Building the Unlimited Version

Our sourcing stack replicates that workflow as an owned pipeline rather than a rented seat.

Data collection runs through scraping infrastructure like Apify, pulling candidate-relevant data at whatever scale the search demands: public profiles, company rosters, job history signals, and the employer-side data that tells you where your candidates are hiding. For a niche manufacturing search, that means mapping every company in the region that employs the skill set, then working inward to the people.

Enrichment and intelligence live in Clay, where an AI research layer does what a tool like Juicebox does internally, except tuned to our market. We write classification prompts that read a company's website and tell us whether it is a precision machining shop or a distributor, which matters enormously when the req demands hands-on plant experience. The same logic scores individual candidates against the actual requirements of the role, applied across thousands of records, with judgment we have encoded ourselves from years of working these searches.

Contact data comes from waterfall enrichment, where multiple providers are tried in sequence for every record until a validated email or phone number comes back. Instead of one vendor's credit allotment, it is a competitive marketplace of data sources, paying only for what resolves. On passive, hard-to-find talent, this is the difference maker. These candidates only respond when you reach them where they actually are, and that requires contact coverage no single metered source provides.

And there is no export step at all, because the pipeline is the system. Candidate records flow from identification to enrichment to multichannel outreach without hitting a download button or a project cap. Every plausible candidate in the market gets found, enriched, and professionally contacted. Total coverage, every search, every time.

The honest tradeoff: this is genuinely harder than signing up for a SaaS seat. It takes engineering, prompt design, data vendor management, and ongoing tending. A solo recruiter working light searches should probably just pay for the tool. But when hard-to-find talent is your bread and butter, the owned pipeline wins on the dimensions that decide searches: cost per validated contact, ceiling on coverage, and how precisely the intelligence layer understands your niche.

Why This Matters

The AI sourcing revolution is real, and it is the best thing to happen to difficult searches in a decade. But the per-seat, credit-metered version of it was designed for individual users running ordinary searches, not for total-coverage sourcing on roles where every uncontacted candidate might have been the placement.

That is the capability we have built: Juicebox-class sourcing intelligence, multi-source data, AI classification, and waterfall contact enrichment, running as dedicated infrastructure with no meter deciding when the coverage stops.

If you have a search that has been open too long, or a niche where the same impossible profile comes up again and again, book a call with us. Bring your toughest req. We will walk through exactly how we would map the market, build the candidate pipeline, and get every findable person in it contacted. That is the fun part.