AI training-data company Mercor uses Serval's IT agent in Slack to onboard 4,000+ contractors automatically, run a 24/7 help desk, and validate engineers' SQL queries before routing them to 14 DBAs for one-click approval

L3 · Conditional Verified · Production Technology IT

Mercor (30,000+ external experts, 250 employees) had IT support running only through a Slack channel with no ticketing and contractor Google accounts created by hand. With Serval, managers start expert onboarding in Slack; Serval checks the requester's permissions, creates a Google Workspace account and applies a role-based access template. Serval answers IT questions around the clock, and engineers post SQL queries in Slack that Serval validates and routes to 14 DBAs to approve with one click, with full audit trails. Workflows are built in natural language with visible generated code, and Serval now spans seven teams. Serval reports 4,000+ contractors onboarded, ticket automation (the source gives both more than 60% of tickets automated and 50% zero-touch resolution, so neither is headlined) and 24/7 coverage (reported by Serval, quoting Mercor Head of IT Dana Stocking; vendor-published)

Problem

No ticketing system, 24/7 support needs across time zones from a small team, hundreds of manual contractor onboarding requests a week, and manual DBA-run database access requests (reported by Serval)

Workflow

  1. Expert manager requests contractor onboarding in Slack
  2. Serval checks requester permissions, creates the Google Workspace account and applies a role-based access template
  3. Employees and contractors get AI answers to IT questions 24/7; unresolved requests become tracked tickets
  4. Engineers post SQL queries in Slack; Serval validates and routes them to DBAs, who approve with one click and results return with an audit trail

Human-in-the-loop: DBAs approve every database query; onboarding is gated by requester-permission checks; IT handles non-automated tickets.

Tools & stack

Tools

  • Serval IT help desk agent
  • Serval natural-language workflow builder

Integrations

  • Slack
  • Google Workspace
  • Production databases

Frameworks

None listed

Prerequisites

  • Slack as the support front door
  • Google Workspace admin APIs
  • Role-based access templates
  • Defined DBA approvers

Risks & limitations

Risks

  • Over-provisioning if access templates are wrong
  • Agent-validated SQL against production data

Limitations

  • Vendor-published; undated
  • Inconsistent automation figure on the page: >60% tickets automated vs 50% zero-touch resolution
  • No time-saved figure quantified

Outcomes & evidence

Reported outcomes

  • 4,000+ contractors onboarded through the automated workflow
    Reported by Serval · Source (serval.com)
  • More than 60% of tickets automated per the page header; the body cites 50% zero-touch resolution
    Reported by Serval · Source (serval.com)
  • Database-access requests that took multiple manual steps now complete in seconds after DBA approval
    Reported by Serval · Source (serval.com)

Evidence

Vendors & implementers

Review history

Last reviewed by Research Chief (agent).

  • : Added in batch-26; source fetched 2026-10-23 and key figures confirmed in page text; figures attributed.
Newsletter

Get new agent use cases in your inbox

A short roundup of newly added and newly verified use cases. Unsubscribe any time.