Metaview is the most accurate AI sourcing coworker in the world. And we have data to prove it.

BLUF: Metaview achieved the highest AI candidate search accuracy in Exa’s People Search Benchmark, scoring 93.5% precision across 1,400 real-world queries.

AI sourcing tools are meant to save recruiters time. In practice, many don’t.

You ask for a Senior Backend Engineer in New York, and get back a long list of almost-right people: wrong location, partial role fit, outdated experience. And a messy understanding of what senior means in your context.

On paper, these tools look like progress. In practice, most just create more work. More profiles to skim, more candidates to reject, more noise flowing into your ATS.

It’s what we call AI sourcing slop: large volumes of mediocre results that feel productive, but quietly burn recruiter time. What you actually want isn’t more candidates. It’s fewer, better results you can trust.

But here’s the thing: not all AI sourcing tools are equal, even if they can be hard to distinguish on the surface.

We put our sourcing agent to the test

Sourcing accuracy can feel subjective. To make it measurable, Metaview’s AI Sourcing agent was evaluated using Exa’s People Search Benchmark, an independent, large-scale test designed to reflect how people actually source candidates.

The benchmark includes:

• 1,400 real-world sourcing queries, each specifying a role, location, and other useful identifiers
• For every query, the top 10 candidates returned are evaluated. Each candidate is checked against the original query to see if they truly match

This isn’t a demo environment full of cherry-picked examples. It’s a broad, repeatable test of whether a sourcing system consistently returns candidates who meet the criteria recruiters asked for. We'll release our test harness on GitHub soon so you can run and verify it yourself.

The results: accuracy that recruiters can actually rely on

On that benchmark, Metaview’s sourcing agent delivered 93.5% precision at scale. Across those 14,000 search results, the vast majority of candidates returned are good fits, not just “close enough.”

By comparison, Juicebox took the same benchmark test and scored 79%. Exa got 63% in its own benchmark, and other AI platforms came out below 30%. That gap in performance may help explain why many recruiters remain cautious about AI sourcing. You can’t possibly automate sourcing if anywhere from 20-70% of results are unusable. 

Why AI sourcing products often disappoint in practice

Many AI sourcing products optimize for volume: returning as many potentially relevant candidates as possible. On a slide, that looks impressive. But in a recruiter’s workflow, it creates slop and results in: 

• More manual filtering
• More time sanity-checking
• Gradual loss of trust in the tool

Products that rely on static filters, keyword matching, or one-shot searches struggle. They don’t know when they’ve failed, and they don’t know how to fix it.

Why Metaview’s sourcing agents are different

Metaview’s AI sourcing agents don’t populate results from simple search filters. They’re agentic, meaning they can reason and reflect on their own output to improve it. Our agents:

Search broadly, then rank aggressively

The agent starts with large candidate pools, but it doesn’t stop there. Before results ever reach a recruiter, candidates are:

• Analyzed
• Triaged
• Ranked against the specific intent of the search

So what you see has already been filtered for quality.

Understand intent, not just keywords

Recruiters don’t always use the exact wording that appears on resumes or LinkedIn profiles. Metaview’s agents understand any form of context you feed it from natural language to voice notes to previous candidates interviews to deeply understand your why you.

Learn over time

Metaview isn’t a single-use tool delivering one-off results. It’s always on, learning from past searches, hiring conversations, and outcomes. Your intake and debrief calls feed into job descriptions and sourcing searches. 

So your entire recruiting process is improved and empowered with AI. And tangibly, you get better at avoiding subpar candidates instead of repeating the same mistakes.

Fewer, but better candidates. Every time.

Accuracy and the data underneath it is our obsession. It’s how we built our notetaker. It’s how we built our sourcing agent.

If results aren’t reliable, AI doesn’t reduce work. It just moves it, and recruiters end up validating, correcting, and second-guessing the output.

When accuracy is high, the dynamic changes. The system improves with feedback, earns trust, and starts to feel like a coworker you can rely on but one that works at a speed no human can.

Ultimately, everything we build points to a simple outcome: helping recruiters and hiring managers hire the right people, as fast as possible. For sourcing, that means tighter shortlists and less time spent filtering.

If you’d like to see how the sourcing agent performs on your roles, try Metaview for free.