Property Management Solution
AI-Powered Account Reconciliation
Property managers reconcile thousands of transactions across operating, trust, and security-deposit accounts every month — bank statements, rent rolls, owner distributions, vendor payables and CAM charges that rarely line up cleanly. Focus20 builds AI agents that autonomously match, investigate, and close every account, turning a multi-week manual grind into a continuous, audit-ready process.
Business Value
A Faster, Cleaner Close
Manual reconciliation is the single biggest bottleneck to owner reporting and month-end close. An AI reconciliation agent works the exceptions the way your best accountant would — reading bank feeds, matching against the ledger, chasing the unexplained variance — so your team reviews a short list of judgement calls instead of scrolling through thousands of transactions.
The Agentic Workflow
1. Agentic Stack
Ingestion Agent: Plaid / bank
feeds + PMS export (AppFolio, Yardi, Buildium).
Matching Agent: Claude
reasoning over ledger + statement lines.
Orchestrator: AWS Step Functions
+ Amazon Bedrock.
2. Reasoning Loop (ReAct)
> Reads memo: "ACH — UNIT 14B TENANT"
> Reasons: "Matches an outstanding May rent receivable, split payment"
> Proposes journal entry + flags for one-click approval
Reconciliation, End to End
Every reconciliation surface a property portfolio touches — handled by AI, reviewed by your team.
Bank Account Reconciliation
Daily auto-match of bank feeds against the ledger across operating, trust and reserve accounts. The agent clears deposits, checks and ACH transfers, and surfaces only genuine exceptions for review.
Trust & Deposit Reconciliation
Continuous three-way tie-out of security-deposit and trust balances against bank and beneficiary ledgers — keeping you compliant with state trust-accounting rules and audit-ready at all times.
Record to Report (R2R)
Automated journal proposals, accruals and intercompany eliminations feeding a faster month-end close — from transaction capture through owner statements and financial reporting.
Rent Roll & AR Matching
Reconciles tenant receipts and partial payments against the rent roll, flags delinquencies, and applies cash to the correct lease and charge code automatically.
AP & Vendor Reconciliation
Matches vendor invoices to payments and statements, detects duplicate or mis-coded bills, and reconciles CAM and recoverable expenses across properties.
Owner Distribution & Draw
Verifies owner-account balances, reserves and management fees before distributions, so every draw ties out to the underlying property ledger.
Things Property Accountants Hear Every Month
Every close comes with the same familiar phrases — and the hours of digging that follow. On the left, what you actually hear. On the right, the agentic AI teammate that ties it out.
“I've sent everything.”
…then three more emails over the week titled “Just one more thing…” Chasing statements is death by a thousand follow-ups.
Keeps a live checklist per property — every bank account, loan and vendor it expects. It spots gaps the moment feeds land and auto-drafts the chaser before you have to.
“The bank reconciles.”
Until you find that mysterious £2.37 that's sat there since 2021 and nobody can explain. Aged items always come back to haunt you.
Never lets an unmatched line quietly age. It surfaces every stale item with its full history, traces the likely origin — “2021 bank fee, never posted” — and proposes the clearing entry.
“It'll only take five minutes.”
Thirty minutes later you're still working out why a loan repayment got posted to “Entertainment.” Miscodes hide in plain sight.
Reads each transaction's memo and counterparty, learns your chart of accounts, and flags the loan-repayment-in-Entertainment before it hits the ledger — with a one-click recode.
“Nothing's changed this year.”
Except the new company, the property purchase, the VAT registration, the payroll, and the three new bank accounts. Everything has changed.
Watches the portfolio for new entities, accounts, properties and tax registrations and quietly onboards them into scope — so surprises arrive as a briefing, not a March scramble.
“Can you just have a quick look?”
Every accountant knows a “quick look” is code for “put the kettle on.” One question unravels into an afternoon across a dozen accounts.
Ask it in plain English — “how does Riverside House look this month?” — for an instant, cited answer: reconciliation status, open exceptions, aged items and cash, drilling down to every source line.
It doesn't replace the accountant's judgement — it hands back the hours lost to chasing and matching, so your team can make the numbers tell the right story, not just the easiest one.
Enterprise Technical Architecture
An AI reconciliation layer that sits on top of the PMS and accounting stack you already run.
AI on Top of Your Ledger
- Data Integration: Bank feeds via Plaid and direct connections, plus PMS/GL exports from AppFolio, Yardi, Buildium, MRI or QuickBooks.
- AI Matching: Claude reasons over statement lines, memos and ledger entries to match fuzzy, split and out-of-order transactions that rules-based tools miss.
- Exception & Audit Trail: Every proposal carries a plain-language rationale and full evidence chain for a defensible audit.
- Human-in-the-Loop: Confidence-scored proposals route to one-click approval; low-confidence items escalate to your controller.
Why AI, Not Just Rules
Legacy auto-match relies on exact rules and breaks on messy memos, partial rent, and lumped ACH batches. Claude reads context the way a human accountant does — resolving the long tail of exceptions that normally consume the close.
Plaid / Direct"] --> ING[Ingestion Layer] PMS["PMS & GL
Yardi / AppFolio"] --> ING ING --> AG[Claude Reconciliation Agent] AG -->|Auto-Match| LEDGER[(Reconciled Ledger)] AG -->|Exception| REV[Controller Review] REV --> LEDGER LEDGER --> RPT[Owner & Financial Reports]
Deployment Timeline
From connected accounts to a continuously reconciled portfolio.
Connect & Map
Connect bank feeds and PMS/GL exports, map your chart of accounts and account types, and baseline against historical statements to learn your portfolio's matching patterns.
Assisted Reconciliation
The agent proposes matches and journal entries with confidence scores. Your team approves or corrects, tuning thresholds and building trust while the model learns your exceptions.
Continuous Close
High-confidence reconciliation runs autonomously every day across the portfolio. Month-end becomes a review of a short exception list, with owner reports and R2R outputs generated on demand.