Talent Intelligence
Underneath every search, score, and hire is one data layer. We call it Manta — 1B+ profiles, 75M companies, 425M+ job postings, always refreshing.
Manta · the engine
Manta is our purpose-built graph engine. The whole product rides on it — sub-second traversal, real-time ingestion, and per-tenant isolation are not features. They’re the substrate.
Sub-second traversal
Custom graph engine. Walks intent → criteria → candidate path in milliseconds, even across hundreds of millions of nodes.
Real-time ingestion
Licensed and public sources refresh hourly. New jobs, new tenures, and new signals land in the graph within minutes.
Tenant-isolated query
Every query runs inside your tenant boundary. Your rubrics, your scores, your hire decisions never inform another customer's matches.
Reasoning trace
Every match comes back with the edges it walked. No black box — you see why this person, not someone else, surfaced for your role.
The knowledge graph
Underneath every rubric, every score, and every hire is the same graph: people connected to skills, signals, companies, and work. Staffer doesn’t “query a database.” It walks a graph that knows where the answer lives.
People
1B+ profiles across 8 public sources, deduplicated to one identity per person.
Jobs
Roles people have actually held — inferred from work history, team context, and trajectory.
Companies
Where people have built. Edge weight changes with role, tenure, and team.
Match path
When a rubric finds a candidate, the walk from intent to person is what lights up.
Staffer is powered by our very own Manta technology.
What the graph knows
Manta indexes the entities that matter for hiring and the relationships between them. Every match is a path; every path is auditable.
1B+
People
Profiles across 8 public + licensed sources, deduplicated to one identity each.
75M+
Companies
Org charts, hiring patterns, headcount trajectories, funding signals.
425M+
Job postings
Roles posted, roles held, transitions, tenure, current openings.
Live
Applications
Every conversation moving through the Platform creates new graph edges.
Continuous
Match paths
The graph recomputes match paths every time a rubric or signal changes.
Layered candidate data
Every candidate is a stack. Hover a row to isolate a single layer, or pick a viewer to see what each role has access to.
The stack
Viewer
Hover a layer or viewer to see what each role has access to.
How the graph learns
Manta sharpens with every action on either side. The Platform labels matches with outcomes; the Portal labels them with consent and correction. Both feed the same graph.
From the Platform
Every action by a hiring team.
Each rubric you write, score you adjust, candidate you advance, and hire you close is a labeled signal. Staffer's intelligence sharpens for your team specifically — without your data ever leaving your tenant.
From the Portal
Every signal from a candidate.
Candidates accept, contest, or decline matches in the Portal. Each interaction tunes the graph — both for their next role and for the matching logic broadly.
Data in. Signals out.
Manta ingests from public sources, your ATS, and live portal activity. Every edge that lands triggers a re-walk — matches, scores, and outreach update in seconds, not overnight.
Trust by default
Your data stays inside your boundary.
Customer data is never used to train shared models. Tenant isolation is the default. EU residency on every plan. Every action is logged, exportable, and reversible.
See it in action
We’ll point Manta at a real opening you have — and you watch the graph walk to the candidates worth meeting.