The VA Group
MAVIS managed AI agent paired with a certified human operator
Businesses wanting a human-supervised AI operations service rather than a conventional standalone virtual assistant.
At a glance
- Support model
- Hybrid AI and dedicated human-in-the-loop support
- Commitment
- 20 or 40 hours weekly
- Starting price
- $2,000
- Price basis
- USD monthly part-time plan; full-time is $3,500 monthly
- Talent geography
- Remote human operators; exact locations require confirmation
What stands out, and what still needs checking.
Clear part-time and full-time prices; human operator; human review; management oversight; integrations; strategy reviews; shared task log; no token or usage overages; broad operational workflows.
The service materially differs from a conventional assistant; buyers need clarity on model vendors, data use and retention, operator location and vetting, approval gates, error liability, security evidence, contract, replacement, and exit portability.
Request architecture and subprocessors, data-retention and training policy, operator profile, access controls, human approval rules, audit logs, incident handling, replacement, term and notice, system ownership, and export support.
What we found and where it came from.
6 public claims, each linked to its source.
Current model1 claim
The current homepage centers on MAVIS, an AI agent operated and supervised by a trained human.
Pricing1 claim
Publishes $2,000 monthly for 20 hours weekly and $3,500 monthly for 40 hours weekly.
Support1 claim
Plans include a certified operator, candidate matching, management oversight, integrations, priority support, and biweekly strategy review.
Hybrid service and pricing1 claim
The VA Group now markets MAVIS, a managed AI agent paired with a certified human operator, at $2,000 monthly for 20 weekly hours or $3,500 for 40 weekly hours.
Included support1 claim
Plans include operator communication during active shifts, hand-selected matching, management oversight, priority support, custom integrations, biweekly strategy reviews, and no token or usage overages.
Buyer diligence1 claim
Public material says a human reviews outcomes before release and work is logged in a shared thread, but contract length, cancellation, replacement, operator geography, vetting, data retention, model providers, access controls, and error liability need confirmation.