01
Same week
Result bound
High-confidence engagements clear without a second reviewer.
For compliance, legal, and ops leaders
Classify every contractor before money moves
Your ops team adds a developer in Germany on Tuesday. By Wednesday, Gruv has assessed the engagement against German labor law, stamped the result with a confidence score and rule version, and blocked payout setup for any employee outcome.
Six-factor assessment per engagementVersioned rule sets (v2024.3, v2026.1)Maker-checker review for borderline callsAutomatic recheck when laws change
Eligible
Contractor, UK design retainer
Independent worker, multiple clients, project-scoped scope, fixed milestones.
CoverageActive
ConfidenceHigh
ReviewerAuto-pass
Review
Dependent, full-time growth lead
Single client, fixed schedule, integrated reporting line, control indicators present.
CoverageHold
ConfidenceMedium
ReviewerMaker-checker
Two engagements, two outcomes. The same questionnaire answered differently produces different coverage, different paper, and different downstream payout posture. AoR keeps that distinction visible.
The classification gate
Decide whether the work fits a contractor model before anything else moves.
Misclassified workers create the most expensive cleanup in any contractor program. AoR runs the assessment against the worker and client context, records the result with confidence and rule version, and routes ambiguous cases through a real review path rather than a one-line admin override.
- Engagement-scoped result, confidence, and rule version on the record.
- Employee, unsupported, and indeterminate outcomes follow stop or review paths.
- Reclassification can mark signed paper stale and trigger a refresh.
Operating tokens
What classification on Gruv looks like
02
6 factors
Per assessment
Behavioral, financial, relationship, scope, jurisdiction, coverage.
03
Reason-coded
Outcomes
Onboarding reads the code. Payouts read the code. Nobody re-asks.
04
12-month
Recheck cadence
Germany updates its rules in March. Gruv queues rechecks in April.
What Gruv records
Six things every classification carries
Every result carries confidence, rule version, jurisdiction, and reason codes on the engagement. The next team acts without re-reading email threads.
Auditors pull one record, not three systems
Result, coverage status, jurisdiction, confidence, and rule version land on the engagement. Your compliance team exports the audit pack in one click.
Onboarding gets a reason code, not a Slack ping
Employee, unsupported, indeterminate, and dependent-contractor outcomes carry reason codes. Onboarding reads the code. Payouts read the code. Nobody re-asks.
Borderline calls go to a second reviewer
A Brazilian engagement scores borderline. Gruv routes it to the maker-checker queue. The second approver, their decision, and the timestamp pin to the record.
v2024.3 reads next to v2026.1
Rule version stamps on every result. When an auditor asks why you classified someone as a contractor two years ago, the answer sits on the same record.
Law changes become a queue, not a fire drill
India updates its labor code in April. Gruv marks affected results stale, queues rechecks, and assigns owners. No calendar reminders to chase.
Every downstream gate reads the same record
Documents, agreements, payout holds, and monitoring all read the engagement record. No team recomputes classification from scratch.
How it works
How a classification moves from input to bound result
Four steps. The result that lands on the engagement is the same record onboarding, agreements, and payouts read downstream.
gruv.app › classification › new engagement
classificationengagementsnew
Step 1 of 4Capture engagement
Engagement nameBrand strategy retainer
WorkerCoral Talent Co.
JurisdictionUnited States
ScopeProject SOW · 12 weeks
Compensation$8,500 / month
Tools provided byWorker
Captured before assessment3 / 3
Engagement context captured
Jurisdiction overlay attached
Ready for assessment
Anatomy of a classification record
Your auditor asks for the result, the data behind it, and the reason it was accepted. Every property below pulls from the engagement in one API call.
Engagement context
Worker country (UK, India, Brazil), client entity, jurisdiction overlay, and engagement type.
Result + confidence
Pass, employee, unsupported, or indeterminate. Each paired with a confidence score.
Rule version
v2024.3 on the original result. v2026.1 on the recheck. Both read side by side.
Reviewer identity
Maker-checker reviewer name and timestamp pinned to every manual outcome.
Recheck cadence
12-month expiry, law-change triggers, and concentration thresholds queue rechecks automatically.
Reason codes
Onboarding reads the code. Payouts read the code. Monitoring reads the code. One source of truth.
Where classification readiness earns its keep
Teams reach for AoR classification when wrong calls show up as legal exposure, payout reversals, or audit findings.
Global contractor programs
You have 200 contractors across the US, UK, Germany, and India. One assessment surface. One record that finance, legal, and ops all read.
Marketplace worker activation
A marketplace worker in an unsupported region never reaches payout setup. Classification blocks activation at the gate.
Compliance-led migrations
Brazil updates its labor code in January. You re-run assessments on your Brazilian cohort and see exactly who needs fresh paper or reclassification.
Audit-ready vendor records
Result, confidence, rule version, and reviewer identity pull in one export. Your auditor stops asking for screenshots from three tools.
After classification
Pass the result into onboarding without losing context
Classification is the first gate. Onboarding collects the W-8BEN, passport, sanctions screening, and payout setup against the same engagement record.
Frequently Asked Questions
How does Gruv pick the right assessment path for a country?+
Gruv reads the worker country, client entity, jurisdiction overlay, and engagement scope. A German engagement runs German labor-law factors. A US engagement runs US factors. The rule version stamps on the result so you can compare Q1 and Q3 assessments side by side.
What happens when a classification call is borderline?+
Indeterminate, dependent-contractor, and manual outcomes route to a maker-checker queue. A second reviewer approves or rejects. The decision and reviewer identity stay on the engagement record.
Does passing classification mean the contractor is active?+
No. Classification is the first gate. The engagement still needs documents (W-8BEN or W-9), a signed agreement, sanctions screening, and payout setup before activation.
What if Germany changes its classification rules?+
Gruv marks affected German engagements stale and queues rechecks with the new rule version. Your compliance team sees exactly which contractors need revalidation and why.
Can we override a classification result?+
Override requests go through a maker-checker path with a second approver. The approval, rejection, and reasoning record against the assessment. Nothing gets lost to a Slack thread.
Ready to map your classification rollout?
Bring your contractor list, launch countries (US, UK, Germany, India, Brazil), and reviewer owners. Gruv scopes the assessments and reason codes your team needs before the first contractor activates.
Many teams start with a narrow launch in weeks.