AI Roadmap · 2026 · Teaser

AI-native
real estate,
from day one.

The industry is shifting from task automation to outcome automation. Propmatic is an agentic operating system for property management — where every workflow runs itself by default, and humans step in only for judgment calls.

21% → 34%
AI adoption among PMs (24→25)
10+ h/wk
Saved per user (early adopters)
72%
Owners investing in AI solutions
81%
Rank data/tech as #1 spend
Market Reality · 2025

The operators who win the next decade
won't be the ones with the best UI.

They'll be the ones whose agents close the most deals, file the fewest tickets, and grow NOI the fastest — without waking anyone up.

KPI
Target with Propmatic
Industry reference
Hours saved per PM / week
10–15 h
AppFolio Realm-X baseline
Lead-to-lease conversion lift
+50–75 %
AppFolio: 73%
AP processing time reduction
50–70 %
Yardi Virtuoso: 60%
Month-end close reduction
70–80 %
Yardi: 20h → 5h
Maintenance resolution speed
30–50 % faster
AppFolio / Yardi data
Lease abstraction time
95 %+ reduction
JLL: $1M hidden revenue found
Architecture

Three layers. One stack.

Every feature on this page fits cleanly into one of these three layers. Nothing ad-hoc.

03

Agentic AI

Do the work

Autonomous multi-step workflows — Leasing Performer, Maintenance dispatcher, Month-end Close, Collections.

02

Generative AI

Help me work

Message drafting, report generation, Ask Propmatic, auto-coded invoices, lease abstraction — LLMs on tap.

01

Predictive / Analytical AI

Tell me what's coming

ML models on structured data — churn risk, dynamic rent pricing, vacancy forecasts, anomaly detection.

🧠
Machine Learning
Predictions from structured data
📝
NLP
Leases, invoices, emails, tickets
👁
Computer Vision
Photos, plans, damage, documents
Generative AI
Drafts, summaries, answers
🤖
Agentic AI
Multi-step autonomous workflows
Competitive Landscape

No single vendor owns the
full AI-native experience.

Yardi has the best data platform. AppFolio has the best agentic leasing. MRI has the best document intelligence. A new entrant designed for agentic workflows end-to-end has a real opening.

MRI Software
Agora / a2i

Embedded across 15+ products

Strengths
  • Lease abstraction (patented OCR)
  • Forms intelligence
  • Ask Agora NLP
Gap
Fragmented; retrofit onto legacy Voyager codebase
AppFolio
Realm-X Performers

Agentic AI for leasing + maintenance

Strengths
  • Image-based maintenance diagnosis
  • 73% lead-to-showing conversion
  • FolioSpace resident UI
Gap
Residential focused; weaker in commercial
Yardi
Virtuoso AI Platform

Marketplace of agents + Composer (no-code builder)

Strengths
  • MCP connectors with Claude
  • Virtuoso Verified invoices (60% AP speedup)
  • Close 20h → 5h
Gap
Requires walled Yardi ecosystem
JLL / CBRE
Custom enterprise AI

In-house models for underwriting & facilities

Strengths
  • CBRE deployed at 20,000+ sites
  • 1B sqft coverage
  • Proprietary data moats
Gap
Not available as SaaS
Bryckel.AI
AI-first lease knowledge base

Pure-play AI lease abstraction

Strengths
  • 90%+ extraction accuracy
  • Fast onboarding
Gap
Narrow scope (leases only)
Six AI Modules · 30+ Features

Every workflow.
One agentic stack.

🏢
Module 01

Property Management AI

Autonomous leasing, pricing, and retention.

1.1

AI-Powered Unit Listings

Auto-generate marketing copy, optimize photo order, SEO-syndicate to Property24 / Zillow / Rightmove, virtual-stage empty units.

TECH · LLM + Computer Vision
Listings go live 10× faster; higher CTR
1.2

Dynamic Rent Pricing Agent

Continuously analyzes comps, seasonality, renewal patterns. Recommends per-unit rent adjustments within guardrails.

TECH · ML regression + market APIs
Optimized revenue; fewer vacancy days
1.3

Leasing Performer (Agentic)

24/7 prospect replies via email/SMS/WhatsApp. Qualifies leads, schedules tours, runs credit/ID checks, pre-fills lease docs.

TECH · Agentic LLM + CRM + DocuSign
+73% lead-to-showing conversion
ref · AppFolio Realm-X benchmark
1.4

Tenant Churn Risk Prediction

Flags at-risk tenants before renewal opens. Signals: unresolved tickets, late payments, portal drop-off.

TECH · ML classification
+5–15% renewal rate
ref · McKinsey 2026 projection
1.5

Vacancy Forecasting

Predicts which units will be empty at 60 / 90 / 120 days. Triggers re-marketing before the unit is actually vacant.

TECH · Time-series ML + churn model
Reduced vacancy loss
💸
Module 02

Accounting & Financials AI

From data entry to autonomous close.

2.1

Invoice OCR + Auto-Coding

Drag-drop invoice → extract vendor, amount, GL code, line items. Auto-match POs, flag discrepancies, label Verified / Warning.

TECH · CV OCR + NLP + ML classifier
3–5× AP throughput, <1% error
ref · Yardi Virtuoso: 60% AP speedup
2.2

Bank Reconciliation Agent

Matches bank lines to ledger. Ranks candidates by confidence. Handles splits, partials, refunds. Learns from corrections.

TECH · ML + fuzzy matching
Monthly rec: days → minutes
2.3

Financial Forecasting

Projects cash flow 30/60/90/180 days. Flags anomalies (utility 3× normal → leak?). Drafts board-ready reports.

TECH · ML time-series + LLM narrative
Fewer cash-flow surprises
2.4

Month-End Close Agent

Coordinates accruals, reconciliations, allocations, JE drafts, CAM rec, straight-line rent. Escalates exceptions.

TECH · Agentic LLM + rules engine
20h → 5h per property
ref · Yardi Virtuoso benchmark
2.5

Collections Agent

Monitors ledgers. Sends graduated reminders in tenant's preferred channel. Drafts settlements. Escalates to legal at the right time.

TECH · Agentic LLM + jurisdictional rules
Lower delinquency, compliant process
2.6

Fraud & Anomaly Detection

Duplicate payments, vendor-on-vendor fraud, synthetic IDs, altered pay stubs, wire-fraud red flags.

TECH · ML anomaly + CV forensics
Reduced loss; regulatory compliance
🛠
Module 03

Maintenance & Facilities AI

Predict, diagnose, dispatch — autonomously.

3.1

Maintenance Performer

Tenant: "AC not cooling." AI asks clarifiers, reads photo, diagnoses, checks warranty, dispatches vendor with parts list, verifies completion.

TECH · Agentic LLM + CV + vendor API
40–60% faster resolution
ref · AppFolio + Yardi pattern
3.2

Predictive Maintenance

IoT sensor data (HVAC, elevator, boiler) predicts failures 7–90 days out. Drone imagery for roof/facade. Lease-clause aware responsibility.

TECH · ML LSTM + CV + MQTT
20–30% maintenance cost cut
ref · Prologis / CBRE data
3.3

Energy Optimization Agent

Ingests smart meters, HVAC, weather. Auto-tunes setpoints by occupancy. Flags hidden leaks. Produces GRESB / ENERGY STAR reports.

TECH · ML + IoT + utility APIs
10–20% utility savings + ESG
3.4

Vendor Intelligence

Ranks vendors by resolution, cost, rework, ratings. Auto-assigns best-fit. Flags price drift. Maintains insurance/licensing compliance.

TECH · ML ranking + OCR for docs
Lower cost, better quality, audit-ready
3.5

Computer Vision Inspection

Move-in/out photo comparison, itemized damage list, deposit deduction report, periodic walk-through video analysis.

TECH · GPT-4V / Claude Vision / Gemini
Consistent deposits, zero missed damage
🧑‍💻
Module 04

Tenant / Resident Portal AI

The 24/7 concierge residents actually love.

4.1

Tenant AI Concierge

Answers any question — rent, lease clauses, pet policy, sublet rules — from their specific lease. Escalates only when it doesn't know.

TECH · LLM + RAG over tenant docs
70–80% resolution without a human
4.2

Smart Rent Payment Assistant

Optimal-time reminders in preferred channel. Voice payments. Personalized payment plans within owner guardrails. Smart waivers.

TECH · LLM + voice AI + ML personalization
Higher on-time rate
4.3

Personalized Recommendations

Learns from portal behavior + surveys. Suggests amenities, events, renewal specials, referral bonuses tailored to each resident.

TECH · Recommendation ML + LLM
Higher engagement, viral referrals
4.4

Sentiment Monitoring

Every interaction analyzed for tone shifts before churn. PM gets alerted with specific recommended intervention.

TECH · NLP classification + alerting
Early-warning CS; better NPS
4.5

Accessibility AI

Real-time translation (30+ languages), voice interface for visually impaired, plain-language lease explanations.

TECH · LLM translation + STT/TTS
ADA / POPIA / GDPR compliant
ref · MRI Agora: 30+ languages
📊
Module 05

Reporting & Analytics AI

Ask Propmatic anything. Ship the answer.

5.1

"Ask Propmatic" — Natural Language BI

"Which buildings will exceed budget next quarter?" → answer + chart + drill-down. "Show tenants whose rent-to-income worsened." → live table.

TECH · LLM + text-to-SQL + viz engine
Entire company queries data without SQL
ref · MRI Ask Agora + Yardi MCP
5.2

Automated Report Generation

Monthly owner reports, investor emails in owner's voice, board-ready slides, ESG reports (GRESB, ENERGY STAR, CSRD) — all auto-drafted.

TECH · LLM + templates + chart rendering
Reporting: days → minutes
5.3

Portfolio Benchmarking

Compares portfolio to anonymized market data. Identifies outliers (NOI lag, above-market OpEx). Recommends specific actions.

TECH · ML clustering + analytics
Data-driven asset strategy
5.4

Scenario Modeling & Digital Twin

What-if: refinance at current rates? Renovate vs divest? Energy upgrade ROI? Fed by real IoT data.

TECH · LLM + financial engine + twin data
Better capital allocation
ref · Gartner: 50% of enterprises by 2028
5.5

Anomaly Alerts

"Utility bill at Building B jumped 45% — possible leak." "Work orders at Unit 7C are 3× normal — tenant may be problematic."

TECH · ML anomaly + alert routing
Problems caught in days, not months
📄
Module 06

Lease Management AI

Turn PDFs into queryable intelligence.

6.1

AI Lease Abstraction

Upload PDF → extract parties, dates, rent, escalations, CAM/NNN, co-tenancy, renewals, termination, use restrictions, default remedies.

TECH · CV OCR + NLP + LLM + fine-tune
95%+ time saved; <1% error
ref · JLL: $1M+ hidden escalations found
6.2

Critical Date & Clause Monitoring

Tracks every renewal, expiry, option, notice period. Auto-reminds at strategic intervals. Flags risky clause combinations (co-tenancy cascades).

TECH · Rules engine + LLM summary
Never miss a date; reduced exposure
6.3

Compliance Engine — ASC 842 / IFRS 16 / GAAP

Auto-classifies operating vs finance. Calculates ROU assets & liabilities. Journal entries. Handles modifications, terminations, impairments.

TECH · Rules engine + LLM reasoning
Audit-ready lease accounting
6.4

Renewal Strategy Agent

Per upcoming renewal: proposes rent, concessions, term. Uses tenant history, market comp, churn score, unit condition. Drafts the offer.

TECH · ML pricing + LLM drafting
Higher retained NOI
6.5

Lease Query AI

"Which leases allow subletting without consent?" "CPI-linked escalations over 5%?" "Co-tenancy tied to Store X?" — answered in seconds.

TECH · LLM RAG over structured leases
Portfolio-wide legal intelligence
6.6

Contract Redlining & Drafting

Compares new lease draft vs standard template, flags deviations, suggests counter-language from playbook. Drafts leases from brief.

TECH · LLM + legal playbook library
Reduced legal cost; faster deals
Cross-Cutting Capabilities

Five primitives.
Every feature uses them.

A
🧩

Agent Marketplace & Composer

Pre-built agents out of the box — Leasing, Maintenance, Collections. Customize via no-code drag-drop. Build your own. Partners publish certified agents to a marketplace.

TECH · Temporal / Inngest + LLM function-calling + MCP
REF · Yardi Virtuoso Marketplace + Composer (Sept 2025)
B
🔌

Model Context Protocol (MCP)

Bring your own LLM — Claude, GPT-4, Gemini, on-prem Llama. Sensitive data stays in Propmatic; only what the LLM needs is exposed. BYOAA workflows.

TECH · MCP server + scoped tool exposure
REF · Anthropic Claude + Yardi Virtuoso Connectors
C
📑

Document Intelligence Platform

Any document — lease, invoice, utility bill, insurance cert, tax form, ID — parsed, extracted into structured tables, and routed to the right workflow.

TECH · Multi-modal LLM + OCR fallback + classification
REF · MRI Forms Intelligence
D
🎙

Voice Interface

PM on-site: "Log work order: Unit 4B, leak under sink, high priority." Tenants complete tasks by voice. Inbound leasing calls handled by AI receptionist 24/7.

TECH · Speech-to-text + LLM + text-to-speech
REF · Mobile-first UX; after-hours coverage
E
🔎

Smart Search Everywhere

"Show me tenants behind on rent with small children." "Find leases expiring in Cape Town with rent over R50,000." Natural language across the whole graph.

TECH · Semantic search (pgvector / Pinecone) + LLM parser
REF · MRI Ask Agora pattern
The Differentiator

Five advantages competitors
can't replicate without rebuilding.

MRI, Yardi, and AppFolio are bolting AI onto software built in the 1990s–2000s. Propmatic is AI-native from day one.

01
Edge 01

Agent-First Architecture

Every feature is built as an agent first, UI second. The UI is for supervision, not execution. Competitors built UIs first and are retrofitting agents behind them.

02
Edge 02

Unified Data Graph

Leases, tenants, units, payments, work orders, and sensors live in one semantic graph — so any agent can reason across domains. Competitors have 10+ DBs with sync issues.

03
Edge 03

Open LLM Layer

Customers bring their own LLM via MCP. No vendor lock-in. AppFolio uses only Amazon Nova. Yardi added Claude recently but is tied to its stack.

04
Edge 04

Autonomous Workflows by Default

Every task has an "autopilot" toggle. Turn it on, the agent runs. Turn it off, it suggests. Competitors' AI is an "assistant" — humans still drive every step.

05
Edge 05

Community-Built Agent Marketplace

Third parties — accountants, brokers, attorneys, regulation experts — publish certified agents. POPIA compliance. Dubai VAT. UK Section 21. Competitors are US-centric.

Positioning Statement

“MRI and Yardi are great accounting tools that added AI chatbots. AppFolio is a great residential PM tool that added agents. Propmatic is an AI-native operating system for real estate where every workflow runs itself by default — and humans step in only for decisions that require judgment.”

PF
Propmatic Manifesto
April 2026 · v1.0
Technical Stack

A stack designed for agents, not apps.

Frontend
  • Next.js 16 (App Router)
  • React Server Components
  • Tailwind + shadcn/ui
API
  • tRPC (type-safe contracts)
  • REST fallback
Agent Orchestration
  • Temporal (workflows)
  • Inngest (events)
  • LangGraph + Vercel AI SDK
  • MCP server
Intelligence Layer
  • LLMs: Claude (primary), GPT-4, Gemini
  • ML Serving: Modal or SageMaker
  • Vector DB: pgvector / Pinecone / Weaviate
Data & Systems
  • Postgres + Prisma (pgvector-enabled)
  • IoT: MQTT / AWS IoT
  • OCR: AWS Textract / Claude Vision
  • Observability: Langfuse / Braintrust
Integrations
  • Auth: Clerk / Auth.js
  • Payments: Stripe + Plaid
  • E-sign: DocuSign API
  • Comms: Twilio · SendGrid · WhatsApp Business
  • Infra: Vercel + AWS/Fly
Six AI Development Principles
01RAG before fine-tuning
02Evals from day one
03Structured outputs only
04Human-in-the-loop by default
05Log every agent decision
06Start narrow, expand gradually
Development Roadmap · 9 Months

Five phases.
One compounding product.

Phase 01Month 1Foundation

Get the non-AI core working so AI has data to work with.

  • Multi-tenant data model (Properties → Units → Leases → Tenants)
  • Rent collection via Stripe/Plaid
  • Basic work order system
  • Tenant portal (web + mobile)
  • Document storage
  • Auth & permissions
Milestone · Core data graph live.
Phase 02Months 2–3Generative AI Layer

Save users time on writing and reading.

  • Ask Propmatic — natural-language BI chat
  • Auto-draft tenant communications
  • AI lease abstraction
  • Invoice OCR & auto-coding
  • Smart search everywhere
  • Auto-generated listing descriptions
Milestone · Users save 5–10 hours / week.
Phase 03Months 4–5Predictive Layer

Anticipate problems before they happen.

  • Tenant churn risk model
  • Dynamic rent pricing
  • Predictive maintenance
  • Financial forecasting
  • Anomaly alerts
  • Vacancy forecasting
Milestone · 5–15% NOI improvement for pilot customers.
Phase 04Months 6-7Agentic Layer

Autonomous workflows that deliver outcomes.

  • Leasing Performer (24/7 leasing agent)
  • Maintenance Performer (triage + dispatch)
  • Collections Agent
  • Month-End Close Agent
  • Renewal Strategy Agent
Milestone · 50%+ of routine PM tasks complete without human involvement.
Phase 05Months 8–9Platform Layer

Become an ecosystem, not just a product.

  • Agent Marketplace (third-party publishers)
  • Composer (no-code agent builder)
  • MCP Connectors (bring your own LLM)
  • Digital twin integration
  • Industry-specific agent packs
Milestone · Network effect — customers choose Propmatic because of its agent ecosystem.
Responsible AI Framework

Table stakes for
enterprise-grade AI.

📋
01

Accountable

Every agent decision is logged with reasoning, inputs, and confidence. Humans can inspect, override, and correct. Developers are responsible for their agents' outcomes.

⚖️
02

Fair & Inclusive

Models tested for bias across protected classes. Tenant screening audited by third parties for disparate impact. No AI can deny housing without human review.

🔍
03

Transparent

Every AI-assisted action is labeled. Users can ask "why did you do that?" and get an explanation. Source documents for AI claims are always linked.

🛡
04

Reliable & Safe

Hard-coded guardrails: no autonomous evictions, no autonomous rent increases above X%, no autonomous tenant rejections. Continuous drift monitoring.

🔐
05

Privacy & Security

Tenant PII isolated per customer. No cross-customer training. SOC 2 Type II. GDPR · POPIA · CCPA · HIPAA. Data residency in EU, US, or ZA.

Jurisdictional Compliance
Global-ready by design
Region
Regulation
Propmatic action
EU
GDPR · EU AI Act
Data residency in EU, right to explanation, bias audits
USA
Fair Housing Act · CCPA
Disparate impact testing, CA privacy rights
South Africa
POPIA
Data residency in ZA, consent management
UK
UK GDPR · Renters Reform Bill
Section 21 compliance agents
Australia
Privacy Act · Residential Tenancy Act
State-specific lease agents
KPIs & Success Metrics

The numbers we commit to.

Every agent ships with an eval. Every customer sees a dashboard. No black-box claims.

Customer-facing · What sales quotes
10–15 h
Saved per PM / week
ref · AppFolio: 10h
+50–75 %
Lead-to-lease lift
ref · AppFolio: 73%
50–70 %
AP time cut
ref · Yardi: 60%
70–80 %
Close time cut
ref · Yardi: 20h → 5h
30–50 %
Faster resolution
ref · AppFolio / Yardi
95 %+
Lease abstraction saved
ref · JLL benchmark
5–15 %
Churn reduction
ref · McKinsey
10–20 %
Energy savings
ref · CBRE · Prologis
3–8 %
NOI improvement
ref · Aggregate impact
Internal · What engineering watches
>90 %
Agent task completion
<10 %
Human override rate (mature)
<1 %
LLM hallucination rate
>95 %
Extraction accuracy
<2 s
Chat latency (p95)
<2 %
Eval regression rate
Closing thought

The next decade in real estate software
belongs to the best autonomous workflows.

MRI, Yardi, and AppFolio have the head start. But they have 20–40 years of legacy code never designed for agentic AI. Propmatic's bet is simple — build the answer from scratch.

Document v1.0 · Updated April 2026 · Prepared for Propmatic MVP