I watched a friend's 18-person recruiting agency evaluate four "AI sourcing platforms" this quarter. All four demoed well. Three got signed. Two got canceled inside 60 days. The fourth is still running but everyone on the team treats it as background noise.
This isn't about any specific vendor. It's about a structural mismatch. Off-the-shelf sourcing AI is built for enterprise buyers — the Randstads, the Robert Halfs — and the product shape that wins those deals makes the product wrong for boutique agencies. Here's where it breaks.
Problem 1: the rubric isn't your rubric
Enterprise sourcing AI scores candidates against a generic fit model trained on millions of LinkedIn profiles and job postings. It's good at "does this person have 5 years of Python." It's terrible at "does this person feel like the weird, scrappy, second-hire at a Series-A that our client actually wants."
Boutique agencies win because their rubric is specific. You know your client's hiring manager personally. You know they'll pass on a Stanford PhD and hire the community-college dropout who's shipped three side projects. A generic model flattens that edge.
Your competitive advantage is a rubric that doesn't exist in any vendor's training data. The moment you adopt theirs, you're a worse Randstad.
Problem 2: the tone is the tell
Open any off-the-shelf outreach AI and read three sequences. You can spot them from orbit. "I hope this message finds you well", "I came across your profile", "impressive background". Candidates spot them too. Reply rates on generic sequences are quietly collapsing — we've seen desks drop from 11% to 3% in 18 months as the LLM-signature became unmistakable.
Your recruiters write in your voice. That voice is why placements renew. A vendor that scales to 10,000 seats cannot customize the voice that matters.
Problem 3: the pricing makes a 20-recruiter desk a rounding error
| Platform type | Seat price | Min seats | Min annual |
|---|---|---|---|
| Enterprise sourcing AI | $250–450/mo | 25–50 | $75k–270k |
| Mid-market "AI recruiter" | $180/mo | 10 | $21.6k |
| Self-serve sourcing tool | $49–79/mo | 1 | $600–950/seat |
| Purpose-built FDE agent | — (retainer) | — | $114k / yr all-in |
The enterprise tier is structurally not for you — the contract alone costs more than your best recruiter's base salary. The self-serve tier is cheap, but it's generic: you get what every other agency gets, which is why nobody wins with it.
Problem 4: the roadmap isn't your roadmap
Every off-the-shelf vendor is optimizing for the next logo on their wall. Your quarterly feedback gets batched with 400 other agencies' into a single backlog ticket. Three months later, you get a Loom announcing a feature that almost does what you asked, but not quite.
A forward-deployed engineer lives in your Slack. You describe a weird edge case on Tuesday; it's shipped Thursday. That velocity is the product.
What to build instead
- Own your rubric. Write your scoring function. A senior engineer can turn your best-placement heuristics into code in a week.
- Own your voice. Tone-match against recordings of your top recruiters, not against generic-professional training data.
- Own your signals. Off-the-shelf tools watch LinkedIn. Your edge might be GitHub, Discord, a specific Substack, the cap-table exits of ten startups you track. That's not a feature any vendor ships.
- Own the roadmap. 2–4 new agents per month, shipped against your next quarter — not the vendor's.
None of this requires 50 engineers. It requires one, embedded, who understands your desk. That's the whole premise of the FDE retainer — and why boutique agencies are, honestly, the best possible fit for it.
If you're currently paying for an off-the-shelf tool that nobody on your team actually uses, book an audit. We'll tell you straight whether an agent would land better, or whether you should just cancel and save the $25k.