The Intelligence Layer

MEL doesn't answer questions.
MEL gets things done.

MEL isn't a chatbot. MEL is an agentic AI with 40+ workflow tools that searches, reports, and takes action across Elevate VMS and Elevate Recruit: with human-in-the-loop safety for high-risk operations.

40+
Workflow tools
7
Personas supported
4
Action categories
Queries on Pro
Four Modes

What MEL actually does.

Search & Report

Find & Summarize

"Show me aging reqs", "candidates in Austin for our React role", "this week's spend by supplier". Answer in seconds, not spreadsheet wrangling.

Take Action

Do the Work

"Approve Sarah's timesheet", "publish this req to Tier 1", "advance the top 3 candidates to interview". One-shot actions with confirmation.

Intake & Create

Generate & Extract

Paste a JD, get a structured req. Paste a resume, get a parsed profile. Run margin calcs from pay rate + skills. Multi-phase intake.

Proactive

Flag & Alert

SLA breaches, aging reqs, expiring credentials, budget overruns: surfaced before they become problems. Contextual, not spammy.

Tool Catalog

40+ workflow tools, grouped by domain.

MEL dispatches to the right tool based on intent. Each tool is a typed, tested API call, not a hallucinated response.

Requisitions (9)

  • • search_requisitions
  • • get_requisition_detail
  • • create_requisition_from_jd
  • • publish_requisition
  • • extend_requisition
  • • close_requisition
  • • list_aging_reqs
  • • margin_calc
  • • taxonomy_lookup

Candidates (8)

  • • search_candidates
  • • parse_resume
  • • score_candidate_fit
  • • compare_candidates
  • • submit_candidate
  • • advance_stage
  • • reject_candidate
  • • schedule_interview

Engagements (6)

  • • list_engagements
  • • extend_engagement
  • • amend_rate
  • • add_task
  • • end_engagement
  • • engagement_summary

Time & Expense (5)

  • • pending_approvals
  • • approve_timesheet
  • • flag_exception
  • • expense_report
  • • bulk_approve

Financials (7)

  • • ready_to_invoice
  • • generate_invoice
  • • invoice_status_change
  • • add_adjustment
  • • tax_override
  • • supplier_payment
  • • spend_report

Analytics & Other (5+)

  • • supplier_scorecard
  • • pipeline_velocity
  • • budget_burn
  • • executive_report
  • • explain_metric
Human-in-the-Loop

AI that knows when to ask.

Every MEL action is classified by risk. Low-risk reads (search, summarize) execute immediately. High-risk writes (approve invoices, close engagements, send customer communications) always propose an action and wait for confirmation.

  • Read-only: executes immediately
  • Medium-risk writes: shows diff, asks to proceed
  • High-risk writes: requires explicit "Yes, do it"
  • Regulated actions: routed to privilege-guarded users
Action Confirmation Flow
HIGH RISK · Customer communication
I'll send invoice INV-2026-0412 ($15,223.40) to billing@meridiancorp.com. Confirm?

See MEL for yourself.

Book a 30-minute demo and watch MEL tackle real workforce tasks: your real questions, your real edge cases, live.