Job Closes → Pull Metrics → Score Vendor → Spot Trends → Update Scorecard

Vendor Performance Agent

Builds a performance record for every vendor automatically after each job: first-trip completion, callback rate, response time, and invoice variance, so vendor decisions run on data instead of the last thing that went wrong.

The Problem

Vendor decisions ride on the last thing that went wrong.

Vendor problems are felt but not tracked. Performance conversations lack evidence.

What Changes

They ride on the data, built after every job.

Summarises response times, repeat issues, and quality indicators from job history. Shows you which vendors are slipping and where - with data, not anecdotes.

Humans stay in control: Your manager decides which vendors to keep and what to pay. The agent assembles the performance data behind those calls. It removes the guesswork, not the relationship.

HOW THE AGENT WORKS

TRIGGERLLMAPIDECISIONACTIONWAIT

TRIGGERJob ClosedA vendor job is marked complete

APIPull Job MetricsGet response time, trips, invoice, callbacks

LLMScore The JobRate against your performance criteria

APIUpdate Vendor RecordAdd this job to the vendor history

LLMSpot TrendsCompare recent jobs to the vendor baseline

DECISIONThreshold Breached?

ACTIONFlag For ReviewSurface a declining vendor

ACTIONRefresh ScorecardPublish the updated scorecard

The agent doesn't pick vendors or set pay. Your manager does. The agent assembles the performance record behind those calls.

HOW TO BUILD IT

Three levels of depth. Pick the one that matches where you are.

For the ops person who needs to understand what's involved and brief their team.

Your JobsClosed vendor work orders

The AgentWorkflow engine + LLM

Your ScorecardLive vendor performance

WHAT YOUR TEAM PROVIDES ONCE

Performance criteria - the metrics you judge vendors on
Job data access - work orders, timestamps, invoices, callbacks
Baselines - what good looks like per metric and trade
Alert thresholds - when a decline is worth flagging
Scorecard format - how you want vendor performance presented

What happens day-to-day: Every time a vendor job closes, the agent records how it went, first-trip completion, response time, any callback, invoice variance. Over a few weeks you have a real scorecard, so when a vendor starts slipping you see it in the data before it costs you a turn.

For the IT person or technical lead who needs to scope the integration.

INQUIRY SOURCES

PMS work orders

AP / invoices

Confirmation logs

WORKFLOW ENGINE (N8N)

Trigger

PMS API - Pull job metrics

LLM - Score job

DB - Update vendor record

LLM - Spot trends

Branch - threshold?

Refresh scorecard

CONNECTS TO

PMS / CMMS (Yardi / Property Meld)Read: work orders, timestamps, callbacks

AP / accountingRead: invoice amounts vs estimates

Vendor history storeWrite: per-job and rolling metrics

Notifications + dashboardFlag declines, publish scorecard

WITHOUT PMS API

Agent works from exported job logs and invoices and updates a scorecard sheet on a schedule. Less timely, but the scoring still runs.

WITH PMS API

Agent pulls real job and invoice data after every close and maintains a live scorecard. Best experience, and it feeds the Confirmation Chaser and No-Show agents.

For the person who wants to build this agent step by step. Each node below maps to an n8n workflow node.

TRIGGERNode 1

Trigger - Job Closed

Webhook when a vendor work order is marked complete.

Output: { job_id, vendor_id, trade, opened_at, closed_at, trips }

APINode 2

HTTP Request - Pull Job Metrics

Gather response time, trip count, callback flag, and invoice vs estimate for the job.

Output: { response_hours, trips, callback, invoice, estimate }

LLMNode 3

LLM - Score The Job

Rate the job against your criteria and baselines for the trade.

Output: { first_trip_complete, response_grade, invoice_variance_pct, job_score }

You score a completed vendor job against performance criteria.

Using the job metrics, the CRITERIA, and the trade BASELINES, return:
- first_trip_complete: true/false (trips == 1 and no callback)
- response_grade: on_time / late, vs baseline
- invoice_variance_pct: (invoice - estimate)/estimate as a percent
- job_score: 0-100 composite per your criteria

Respond ONLY with valid JSON.

CRITERIA: {criteria}
BASELINES: {baselines}
JOB: {job_metrics}

INPUT

{ "response_hours": 6, "trips": 1, "callback": false, "invoice": 285, "estimate": 250 }

OUTPUT

{
  "first_trip_complete": true,
  "response_grade": "on_time",
  "invoice_variance_pct": 14.0,
  "job_score": 88
}

APINode 4

HTTP Request - Update Vendor Record

Append the job score and metrics to the vendor’s rolling history.

Output: Vendor history updated with this job.

LLMNode 5

LLM - Spot Trends

Compare the vendor’s recent jobs to their baseline and flag meaningful decline.

Output: { trend, flagged_metric, summary }

You detect performance trends for a vendor.

Given the vendor's recent job scores and their baseline, return:
- trend: improving / stable / declining
- flagged_metric: the metric driving a decline, or null
- summary: one line for a manager

Only flag declining if the change is consistent, not a single bad job.
Respond ONLY with valid JSON.

RECENT JOBS: {recent_jobs}
BASELINE: {baseline}

OUTPUT

{
  "trend": "declining",
  "flagged_metric": "first_trip_completion",
  "summary": "First-trip completion fell from 90% to 60% over the last 5 jobs."
}

DECISIONNode 6

Condition - Threshold Breached?

IF a flagged decline crosses your alert threshold → flag for review. ELSE → just refresh the scorecard.

Output: Branch: BREACH (flag) or OK (refresh only)

ACTIONNode 7

Refresh Scorecard & Flag

Update the published scorecard; if breached, notify the manager with the trend summary.

Output: Scorecard current; declining vendors surfaced.

MVP vs Advanced — Where To Start

MVP — START HERE

Manual job-close trigger
Core metrics (trips, response)
Static baselines
Per-job scoring
Scorecard on a sheet
Decline flag by email

ADVANCED — ADD LATER

Full Video Walkthrough Coming Soon

From closed jobs to a live vendor scorecard - step by step.

Everything on this page is free to learn.

When you're ready to build — do it yourself or bring us in.

Book a call to scope this

OTHER AGENTS IN LEASING & MARKETING

COI Gate Agent

Blocks dispatch when compliance docs are missing or expired

Vendor No-Show and Recovery Agent

Detects missed windows, triggers rescheduling and backup outreach

Vendor and Contract Renewal Timeline Agent

Flags renewals early so terms get negotiated on time

Bid Normalisation Agent

Converts proposals into comparable scopes, highlights gaps

Frequently asked questions

Vendor Performance Agent is an AI agent that builds performance profiles from response times and job history.

Vendor decisions ride on the last thing that went wrong. The Vendor Performance Agent addresses this — it builds performance profiles from response times and job history.

No — your team stays in control. Your manager decides which vendors to keep and what to pay. The agent assembles the performance data behind those calls. It removes the guesswork, not the relationship.

This is 7 of 7 agents in Vendor Management. → · Explore all 10 departments →