Reference case study
The Dominion Group
Xandec worked across The Dominion Group, from financial services to property operations, to turn AI from a leadership priority into production systems and employee workflows.
Deployment scope
2
Business lines
10+
Portals unified
1,000+
Units managed, supplied figure
Portfolio figure supplied for this case study. Final publication should reconcile units managed with the broader Helm property registry.
The starting point
The Dominion Group wanted to become an AI-enabled operating company. That meant finding where work slowed down across both business lines, building reliable data foundations, shipping useful systems, and teaching people how to use AI inside the work they already owned.
Dominion Financial Services
Staff verified borrowers, reviewed mortgage information, collected documents, calculated DSCR loans, priced deals, and moved cases through portal workflows with too many manual steps.
The issue was not one missing screen. It was the cumulative delay created by repeated checks, handoffs, and systems that did not give borrowers or staff a clear path through the process.
Dominion property operations
AppFolio ran day-to-day property management. Buildertrend ran construction. Podio held acquisition, renovation, leasing, and compliance records. QuickBooks supported accounting. Teams maintained mirrors and reconciled handoffs by hand.
Every property moved through acquisition, renovation, leasing, active rental, and turnover. A missing or stale record could hide delay, create duplicate entry, or route a decision using the wrong owner or status.
Business line one: financial services
The financial services work focused on the places where manual verification, calculations, document movement, and limited portal experiences added friction. Each product removed a specific part of that burden while preserving human review where it mattered.
01
Operational problem
Pricing a loan required staff to collect deal inputs, check program rules, and calculate terms before a borrower could get an answer.
What changed
A live pricing workflow turns structured deal inputs into an immediate pricing path for borrowers and the team.
The workflow organizes property, borrower, loan-purpose, leverage, and pricing inputs before a team member reviews the proposed terms.
02
Operational problem
Identity checks, mortgage verification, borrower documents, and DSCR loan calculations moved through manual handoffs and disconnected records.
What changed
The borrower portal creates one guided intake and verification surface for the lending workflow.
The review path covers borrower identity, entity information, property details, mortgage information, and supporting documents before underwriting review.
03
Operational problem
Legacy portal workflows, including LoanPASS, slowed document exchange and made status harder for clients and staff to follow.
What changed
The client portal gives customers a clearer place to submit information, track progress, and continue the process.
The new experience is designed around clear requests, visible progress, and fewer off-platform document and status handoffs.
04
Operational problem
As the number of production portals grew, employees and customers needed one place to confirm whether each service was available and healthy.
What changed
A public status surface monitors the pricing, borrower, and client portals and makes current service health visible without a support request.
The status layer turns portal health into a shared operational signal and gives the team a clear place to communicate service availability.
Business line two: property operations
Property operations were not a linear software problem. A property crosses legal entities, departments, and systems from contract through renovation, leasing, active rental, and turnover. We documented that lifecycle, the people responsible for each transition, and the source of truth at each stage.
The operating principle
Every day a property is owned but not rented carries a cost.
Helm uses one lifecycle model to show where those days accumulate. Phase clocks locate delay. Cross-system rules expose the records that make the lifecycle unreliable.
The foundation: Helm
Helm consolidates the signals needed to understand a property across its lifecycle. PostgreSQL provides the shared data foundation. The pipeline reconciles identifiers, derives lifecycle phases, and applies audit rules across operational records. Claude supports extraction and classification where the source material is unstructured.
Property operations architecture
Operational sources stay accountable for their part of the lifecycle. Helm creates the shared view across them.
Property management
Acquisition and compliance
Construction
Accounting context
Shared foundation
Normalized properties, units, owners, jobs, source evidence, and lifecycle snapshots.
Claude-assisted pipeline
Extraction and classification for maintenance notes, scopes, invoices, and other operational text that arrives without a reliable schema.
One property spine across systems, owners, units, jobs, and lifecycle evidence.
A phase model locates every property and exposes stalled or incomplete handoffs.
Rules flag owner, status, rent, licensing, certificate, and turnover discrepancies for review.
AI in the hands of employees
Production systems created leverage at the company level. Adoption required a second layer: putting AI into the daily work of the people who price loans, review documents, lease units, run construction, reconcile records, and manage compliance.
Before
Manual verification, document review, pricing, and DSCR calculations across several steps.
With the deployment
AI-assisted workflows inside pricing, borrower intake, and internal review.
AI prepares and structures the work. Lending staff retain responsibility for verification, exceptions, and final decisions.
Before
AppFolio held the live leasing workflow while staff and assistants mirrored status into Podio.
With the deployment
Helm creates a shared property view and exposes lifecycle exceptions without another manual lookup.
A leasing or operations user can review one property record with source evidence and exceptions instead of rebuilding context across platforms.
Before
Project managers worked in Buildertrend while administrators maintained a separate Podio renovation mirror.
With the deployment
The unified layer connects construction status to the property lifecycle and the handoff back to leasing.
Construction status and handoff evidence are normalized so project and property teams can identify missing steps before they become hidden delays.
Before
Teams cross-checked owner entities, rent, licenses, certificates, invoices, and operational records by hand.
With the deployment
Cross-system audit rules surface mismatches and expiring records in one place for review.
The system structures supporting records and directs mismatches to a human reviewer rather than making autonomous accounting or compliance decisions.
Forward deployed engineering and GTM
This engagement combined two kinds of deployment. Forward deployed engineering embedded technical builders inside the organization to understand and improve how work moved. The GTM track applied the same systems approach to finding, enriching, and qualifying new leads.
Forward deployed engineering
As forward deployed engineers, we did more than take software requirements. We worked alongside lending, leasing, construction, property operations, and accounting to understand the decisions, handoffs, records, and exceptions that shaped daily work.
Map
Documented business lines, roles, systems of record, and property lifecycle handoffs.
Build
Connected fragmented data and shipped production tools around real employee workflows.
Deploy
Put AI into the work itself, with domain owners involved in testing and review.
Train
Developed the playbook for training FDEs and enabling employees to continue using AI effectively.
A representative FDE workflow connected property source data, lifecycle evidence, and audit exceptions so operations staff could investigate a property without rebuilding its history by hand.
Embedded business functions
01 Map
Roles, ownership, handoffs, approval boundaries, and daily workflows.
02 Foundation
Normalize system records and structure operational text into shared context.
03 Surfaces
Deploy production tools into the workflows employees already own.
AI inside employee workflows
Employees use structured context, AI assistance, and visible exceptions with human decision ownership preserved.
Go-to-market engineering
For the leasing GTM workflow, we built a waterfall model that sources potential leads from CoStar, then passes each record through multiple research, extraction, and enrichment services before it reaches the team.
Source 01
Commercial property and market lead universe.
Source 02
Property, ownership, and public-record context.
Source 03
Generate and expand target candidates from defined criteria.
01 Research
Verify company, property, and market context against current sources.
02 Extract
Collect websites, contact data, and supporting public evidence.
03 Resolve
Normalize records, remove weak matches, and resolve duplicate entities.
04 Qualify
Apply leasing criteria and send qualified records to the correct motion.
Destination
Review qualified accounts with source evidence attached.
Communication
Support structured calling, messaging, and follow-up workflows.
Learning loop
Return outcomes to targeting and scoring criteria for the next run.
CoStar provides the starting universe. Claude and Perplexity support research and qualification. Apify and additional public-record and contact-enrichment services collect and normalize the evidence used to clean each lead.
The model
The same delivery discipline connected both tracks: understand the operating context, build against real constraints, deploy with the people doing the work, and leave behind a capability that can be repeated.
Why Claude
Technical stack
AI and data
Products
Business systems
GTM and communications
Engineering
Helm is the property operations product. HubSync is the synchronization and normalization layer supporting its shared property view. LoanPASS remains part of the financial-services system landscape while the newer pricing and portal surfaces improve the borrower and client experience around it.
Deployment scale
These figures describe the documented scope of the deployment. Performance outcomes such as hours saved, conversion lift, and cycle-time reduction should be added only after Dominion validates the measurement.
2
Financial services and property operations
10+
Customer and internal operating surfaces
5+
Pricing, borrower, client, status, and supporting experiences
1,000+
Residential rental portfolio
Build the next deployment