REGISTER REGISTER TO LEARN

Pull Portfolio Data → Detect Patterns → Rank Hotspots → Explain → Report

Portfolio Insights Agent

Scans the whole portfolio for the patterns a single property view hides: repeat failures, backlog spikes, cost anomalies, and vendor concentration, and surfaces the few that deserve action this week.

The Problem

Forty properties. Nobody has time to compare them.

Portfolio-level patterns - repeat failures, cost spikes, backlog trends - stay invisible until they become crises.

What Changes

The few patterns that matter, surfaced each week.

Flags hotspots across properties using operational data. Shows you which property is trending toward a problem before the problem arrives.

Humans stay in control: Your leadership decides what to act on. The agent finds and ranks the patterns. It removes the digging, not the decisions.

HOW THE AGENT WORKS

TRIGGERLLMAPIDECISIONACTIONWAIT
TRIGGERScheduled RunWeekly portfolio scan
APIPull Portfolio DataTickets, turns, costs, vendors across properties
LLMDetect PatternsFind spikes, repeats, anomalies
LLMRank HotspotsOrder by impact and trend
DECISIONWorth Flagging?
LLMExplain EachWrite the why and the suggested look
ACTIONSend ReportDeliver the ranked insight brief
The agent doesn't decide what to act on. Leadership does. The agent finds and ranks the patterns worth a look.

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 PortfolioAll properties, all systems
The AgentWorkflow engine + LLM
Your BriefRanked hotspots

WHAT YOUR TEAM PROVIDES ONCE

  • Data access - tickets, turns, costs, vendors across the portfolio
  • Pattern definitions - what a spike, repeat, or anomaly means
  • Baselines - normal ranges per metric and property type
  • Ranking priorities - what makes a hotspot worth attention
  • Report format - how leadership wants insights presented
What happens day-to-day: No one has time to compare 40 properties by hand. Each week the agent scans them all, finds the three or four real patterns, a backlog spike here, a repeat-failure cluster there, a vendor running hot on cost, and explains each one. Leadership reads a short brief instead of mining dashboards.

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

INQUIRY SOURCES

PMS / CMMS
AP / accounting
Vendor + turn data

WORKFLOW ENGINE (N8N)

Trigger - schedule
API - Pull portfolio data
LLM - Detect patterns
LLM - Rank hotspots
Branch - worth flagging?
LLM - Explain each
Send report

CONNECTS TO

PMS / CMMS (Yardi / AppFolio / Property Meld)Read: tickets, turns, backlog
AP / accountingRead: cost data by property and category
Vendor + turn agentsRead: performance and turn-risk signals
Delivery + dashboard (Slack / BI)Send brief, feed an exec view
WITHOUT PMS API

Agent works from exported reports per property and produces the brief from those. Less timely, but the pattern detection still runs.

WITH PMS API

Agent reads every system across the portfolio and produces a live weekly insight brief. Best experience, and a capstone over the operational 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 - Scheduled Run
Cron weekly (e.g. Monday 6am) across the full portfolio.
Output: { run_date, properties[] }
APINode 2
HTTP Request - Pull Portfolio Data
Parallel GETs for tickets, turn status, costs, and vendor metrics across all properties, normalised per property.
Output: { per_property: { tickets, backlog, turn_days, cost_by_cat, vendor_mix } }
LLMNode 3
LLM - Detect Patterns
Compare each property to its baseline and the portfolio median to find spikes, repeats, and anomalies.
Output: [{ property, pattern, metric, value, baseline, severity }]
You detect operational patterns across a property portfolio.

For each property, using BASELINES and PATTERN DEFINITIONS, find notable patterns:
- pattern: backlog_spike / repeat_failure_cluster / cost_anomaly / vendor_concentration / turn_slippage
- metric, value, baseline: the numbers behind it
- severity: high / medium / low

Only return real deviations, not noise. Respond ONLY with a valid JSON array.

BASELINES: {baselines}
PATTERN DEFINITIONS: {pattern_definitions}
PORTFOLIO DATA: {portfolio_data}
OUTPUT
[
  { "property": "Maple Court", "pattern": "backlog_spike", "metric": "open_wo", "value": 42, "baseline": 18, "severity": "high" },
  { "property": "Oak Ridge", "pattern": "cost_anomaly", "metric": "plumbing_spend", "value": 9800, "baseline": 4200, "severity": "high" }
]
LLMNode 4
LLM - Rank Hotspots
Order the detected patterns by impact and trend; keep only the few worth leadership attention.
Output: [{ rank, property, pattern, impact, trend }]
You rank portfolio hotspots for a weekly leadership brief.

Order the detected patterns by impact (severity x scope) and trend (worsening first). Return the top items only, as a JSON array with rank, property, pattern, impact, trend.

PATTERNS: {patterns_json}
OUTPUT
[
  { "rank": 1, "property": "Oak Ridge", "pattern": "cost_anomaly", "impact": "high", "trend": "worsening" },
  { "rank": 2, "property": "Maple Court", "pattern": "backlog_spike", "impact": "high", "trend": "new" }
]
DECISIONNode 5
Condition - Worth Flagging?
IF ranked hotspots exist above the attention threshold → explain and report. ELSE → send an all-clear summary.
Output: Branch: HOTSPOTS (explain) or QUIET (all-clear)
LLMNode 6
LLM - Explain Each Hotspot
For each top hotspot, write a short why and a suggested first look, in plain language.
Output: [{ property, headline, why, suggested_look }]
You explain portfolio hotspots for leadership.

For each ranked hotspot, write:
- headline: one line
- why: the likely driver, grounded in the data (no speculation beyond it)
- suggested_look: the first thing to check

Plain language, no jargon, no recommendations you can't support. Respond ONLY with a valid JSON array.

HOTSPOTS: {ranked_hotspots}
SUPPORTING DATA: {supporting_data}
OUTPUT
[
  { "property": "Oak Ridge", "headline": "Plumbing spend more than doubled vs baseline", "why": "Three high-cost emergency calls to one unit suggest a recurring failure being patched repeatedly.", "suggested_look": "Pull WO history for the unit; consider a root-cause repair." }
]
ACTIONNode 7
Send Report
Format the ranked, explained hotspots into a short brief and deliver to leadership; feed the exec dashboard.
Output: Weekly portfolio insight brief delivered.
MVP vs Advanced — Where To Start
MVP — START HERE
  • Manual run trigger
  • One system (tickets)
  • Fixed baselines
  • Simple spike detection
  • Single brief
  • Delivered by email
ADVANCED — ADD LATER

    Full Video Walkthrough Coming Soon

    From portfolio noise to ranked hotspots - step by step.

    Register to be notified

    Everything on this page is free to learn.

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

    Register FOR WALKTHROUGHS
    Book a call to scope this

    OTHER AGENTS IN LEASING & MARKETING

    Executive Summary and What Changed Reporting Agent

    Generates weekly/monthly narratives from KPIs and exceptions

    Daily Exceptions Brief Agent

    Morning list of what's stuck, who owns it, what's next

    Owner Reporting Agent

    Drafts owner-ready summaries with variance explanations

    Frequently asked questions

    Portfolio Insights Agent is an AI agent that flags hotspots - repeat failures, cost spikes, and backlog trends.
    Forty properties. Nobody has time to compare them. The Portfolio Insights Agent addresses this — it flags hotspots - repeat failures, cost spikes, and backlog trends.
    No — your team stays in control. Your leadership decides what to act on. The agent finds and ranks the patterns. It removes the digging, not the decisions.

    This is 4 of 4 agents in Executive & Portfolio Operations. → · Explore all 10 departments →