Table of Contents

1. Key Highlights
2. Introduction
3. From a failed experiment to a deliberate strategy: how the early tests reshaped priorities
4. Replacing background work, not relationships: design principles for agent-centered AI
5. Reconnecting quiet clients: how smarter data makes outreach feel personal again
6. Personalizing the luxury experience: merging Christie’s auction data with real estate relationships
7. The technical backbone: how these features work without mystifying agents
8. Ethics, privacy, and brand authenticity: managing risks when AI touches client data
9. Training agents and reshaping workflows: adoption is as much cultural as technical
10. Practical adoption playbook: step-by-step for brokerages of any size
11. Measuring impact: the KPIs that matter
12. Common pitfalls and how to avoid them
13. How smaller brokerages can get started without a full AI team
14. The role of leadership: incentives, accountability, and long-term thinking
15. Where the industry is headed: agent roles, hybrid workflows, and the next decade
16. Real-world signals: acquisitions, partnerships, and market context
17. Practical example: what an AI-assisted luxury touchpoint looks like
18. Building trust with clients: messaging and consent strategies
19. The economics: where the value accrues
20. What to ask a vendor or partner: procurement checklist
21. The competitive edge: why authenticity beats mimicry
22. Looking ahead: practical signals to watch for industry-wide change
23. FAQ

Key Highlights

• @properties Christie’s International Real Estate shifted from trying to replace agent-client conversations with AI to automating background tasks and enriching personal connection, boosting agent productivity and client engagement.
• Deep personalization—linking client interests across art, wine, and luxury goods with Christie’s auction data—creates meaningful outreach opportunities and a scalable luxury experience without replacing the human agent.
• Successful AI adoption requires clear design principles, careful data handling, agent training, and metrics focused on relationship value rather than automation for its own sake.

Introduction

Real estate depends on relationships. Agents win business by combining market knowledge with trust, timing, and a distinctive personal touch. When Thad Wong began building AI for @properties Christie’s International Real Estate, the first instinct was to automate conversations and scale human interaction. That experiment failed in a revealing way: clients rejected synthetic substitutes for authentic agent voice. The breakthrough didn’t come from teaching machines to impersonate real estate professionals; it came from teaching machines to remove friction—repeating tasks, research chores, and outreach preparation—so agents could spend more time on high-value, human activities.

This article traces the evolution of Wong’s approach, explains how AI augments rather than replaces agents, unpacks the luxury-focused personalization enabled by the Christie’s relationship, and sets out practical guidance for brokerages that want to replicate those gains without eroding client trust or running afoul of privacy expectations.

From a failed experiment to a deliberate strategy: how the early tests reshaped priorities

Early prototypes attempted to have AI stand in for agents during initial conversations, maintaining a dialogue until a substantive moment triggered a handoff. That model was attractive on paper: scale human-like engagement at low marginal cost. In practice it undermined the core value agents provide.

Agents deliver referrals and repeat business because of two intertwined assets: a personal identity that clients trust, and a context-rich understanding of client needs spanning family, tastes, and life plans. When AI impersonates those qualities, the relationship devalues. Wong described the effort as “a really bad idea” that taught a crucial lesson: technology must amplify what makes an agent valuable rather than supplant it.

That lesson reframed the project. The team’s mandate shifted from replicating conversations to eliminating “unseen and repeated behaviors” that consume agents’ time: manual research, administrivia, spotty follow-up, and superficial mass-communication. Free those minutes and agents can concentrate on tailored conversations and complex negotiations—the parts of the job that produce real client loyalty and referrals.

The pivot illustrates a general rule for AI in services: automation succeeds when it augments the unique human contribution instead of faking it.

Replacing background work, not relationships: design principles for agent-centered AI

Design choices that prioritize relationship preservation produce better outcomes. Wong’s team followed multiple practical principles that apply across brokerages:

• Preserve identity control. Agents must own how they present themselves. Any messaging or automation that sounds like an agent must be explicitly approved and reproducible by that agent. Sending AI-generated “voices” that impersonate an agent without oversight breaks trust.
• Automate the repetitive, surface-level tasks. Pulling property histories, generating comparative market analyses, scheduling, and populating contact notes are low-value, high-effort duties where AI and automation deliver immediate ROI.
• Enhance context for conversations. Tools that synthesize a client’s past interactions, personal interests, and public social posts provide agents with conversation starters that feel authentic. A well-timed reference to a client’s hobby or prior comment creates rapport without resorting to imitation.
• Recommend action with rationale. Prioritization systems should not only identify who to contact but explain why and suggest the content of the outreach. Agents respond when the system connects actions to clear outcomes, such as an increased likelihood of conversion.
• Build guardrails for authenticity. Templates, tone recommendations, and content snippets help agents start conversations more quickly while preserving their unique voice.

These principles turned a failed conversational experiment into a coherent toolset that extends agents’ reach and preserves the human element.

Reconnecting quiet clients: how smarter data makes outreach feel personal again

A recurring problem for agents is how to re-engage past clients without sounding transactional. Cold, generic “checking in” emails underperform; agents worry about appearing opportunistic. AI can bridge that gap by turning forgotten details into meaningful prompts.

Wong’s team built tools that ingest historical communications and social signals—email threads and social media posts from platforms such as Facebook, Instagram, and LinkedIn—then surface memorable hooks from those records. An agent preparing to call a former client might be shown a timeline of past conversations, a note that the client just posted images from a family trip, or that they follow a particular artist. These tailored prompts make the outreach relevant and personable.

Illustrative scenario

• An agent sees that a former client recently posted photos from a vineyard trip and had referenced enjoying Italian wines in a message a year prior. The AI prepares a short outreach script: congratulate them on the trip, link to a regional wine-tasting event, and mention a relevant new listing with a suitable cellar. The client responds, conversation resumes, and the relationship rekindles without any sense of automated intrusion.

The system also prioritizes outreach lists, scoring contacts by likelihood of responsiveness or value and explaining the signal behind each recommendation. Agents receive a short “why reach out” note—for example, “High propensity to move: homeowner in neighborhood with rising listings; engaged on social with recent home-improvement posts”—and a suggested opening. This targeted approach increases acceptance of outreach and the efficiency of follow-up.

Personalizing the luxury experience: merging Christie’s auction data with real estate relationships

The partnership between Christie’s International Real Estate and Christie’s auction house offers a rare cross-domain dataset: property buyers who are also collectors of art, watches, handbags, and wine. Wong’s team designed systems to exploit that overlap for deeper, meaningful engagement.

Instead of proclaiming Christie’s as a luxury brand, the brokerage uses client-specific interest data to generate timely, relevant opportunities. If a client follows a certain contemporary artist, agents can be alerted when the artist’s work appears in an upcoming auction and prompted to share the news with a personalized message and auction link. That outreach strengthens the agent’s relevance between property transactions and introduces high-potential buyers to Christie’s auction offerings.

This strategy creates mutual value:

• The auction house gains access to a curated audience of potential buyers who have verified interest in similar categories.
• Real estate agents gain a reason to contact clients that has nothing to do with selling a house, increasing the frequency and quality of interactions.

Scaling this personalization requires robust identity resolution and consent-aware data practices. Aggregating social signals, CRM notes, and purchase or browsing behaviors across Christie’s platforms creates profiles that inform agent outreach at scale. The result is a luxury experience tailored to each client’s tastes without substituting a human touch.

The technical backbone: how these features work without mystifying agents

The features Wong describes sound sophisticated, but the architecture behind them follows current industry patterns. High-level components include:

• Unified CRM ingestion. Consolidating email archives, transaction histories, and public social data into a single profile enables rapid retrieval of context.
• Natural language understanding (NLU). NLU models extract entities and sentiments from prior conversations and social posts for conversational prompts—names, dates, favorite artists, or references to a child’s graduation.
• Vector embeddings and semantic search. Embeddings turn unstructured text into searchable vectors so the system can find semantically similar past interactions or client signals, rather than relying on brittle keyword matches.
• Personalization models and propensity scoring. Machine learning models predict which past clients are most likely to respond or transact and surface the top reasons. These models use features like recency of engagement, transaction history, local market trends, and social activity.
• Retrieval-augmented generation (RAG) for content drafting. Where agents want help with message phrasing, RAG lets the system generate suggested outreach drafts grounded in the client’s profile and agent-approved templates.
• Audit and approval workflows. Agents review and edit suggested messages, ensuring human control and compliance with brand voice.
• Monitoring and KPIs. Dashboards track re-engagement rates, conversions attributed to AI-suggested outreach, agent time saved, and client satisfaction metrics.

This stack emphasizes human-in-the-loop operation. Agents receive high-quality inputs and retain final decision-making power. The architecture balances automation with transparency.

Ethics, privacy, and brand authenticity: managing risks when AI touches client data

Collecting and analyzing client communications and social signals creates ethical and legal obligations that must be addressed through design and policy.

Consent and transparency

• Obtain informed consent for using private communications. Importing archived emails or messages requires explicit authorization. Public social posts are different legally, but transparency about how public signals will be used builds trust.
• Allow opt-outs. Clients who prefer lower touch or tighter privacy should be able to limit the types of personalization applied to them.

Authenticity and impersonation

• Never automate messages so that they mimic an agent without disclosure. Agents must approve messages, and systems should avoid synthetic voice or text that claims to be the agent unless explicitly permitted and controlled.
• Maintain audit trails. Every AI-generated suggestion and the agent’s final version should be logged for accountability.

Bias and profiling

• Propensity models can inadvertently reinforce biased assumptions. Regularly audit models for unfair treatment—particularly for price appraisal, target outreach, and lead scoring.
• Use demographic features cautiously and only when justified by transparent performance measures.

Security and data protection

• Store sensitive client data with strong encryption and limit access to necessary personnel.
• Comply with local regulations such as CCPA, GDPR, and other privacy laws that affect data handling and portability.

These safeguards protect clients, preserve brand integrity, and prevent the worst outcomes—alienated clients, regulatory fines, or reputational damage.

Training agents and reshaping workflows: adoption is as much cultural as technical

Technology alone does not change behavior. Adoption requires reshaping daily routines, measuring new KPIs, and equipping agents with both skill and the confidence to use AI responsibly.

Training programs should include:

• Hands-on sessions that let agents practice editing AI-generated drafts and see how personalization features improve response rates.
• Role-based tutorials illustrating when to use automation—scheduling and research—and when to lead with a human touch, such as negotiation or emotional responsiveness.
• Playbooks that map AI outputs to concrete actions: e.g., “If the system flags a contact as high propensity because of a recent social post about moving, call within 48 hours and mention X.”

Change management also requires executive sponsorship. Leadership must demonstrate that AI’s purpose is to increase agent effectiveness, not to reduce headcount. Communicate success stories: agents who re-engaged clients because AI surfaced a poignant detail, or who shortened time-to-listing by automating paperwork.

Measure the right outcomes

• Time reallocated from administrative tasks to client-facing activity.
• Re-engagement rate among dormant contacts.
• Conversion lift attributable to AI-assisted outreach.
• Net promoter score (NPS) changes and retention among clients who receive AI-personalized interactions.

Adoption scales when agents see direct improvements in productivity and revenue tied to technology use.

Practical adoption playbook: step-by-step for brokerages of any size

Brokerages with different resource levels can apply Wong’s lessons. The approach varies depending on scale, but the core sequence remains consistent.

Step 1: Define the human value you want to protect and augment Identify the agent behaviors that create the most client value—personal storytelling, negotiation, local market advising—and design AI to protect and extend them.

Step 2: Start small with high-impact automation Automate a single low-risk task: CRM cleanup, contact prioritization, or automated property-history generation. Measure time saved and agent satisfaction.

Step 3: Build consent-forward data collection Invite agents and clients to opt into enhanced personalization features. Make it simple for clients to understand and control what data is used.

Step 4: Deliver contextual prompts, not scripts Provide agents with context-rich prompts and suggested openings; avoid fully autonomous messaging that bypasses the agent’s voice.

Step 5: Iterate on models and workflows Use early data to refine propensity scoring and personalization rules. Track false positives—contacts flagged incorrectly—and adjust thresholds.

Step 6: Scale toward richer personalization Once basic features prove ROI, integrate external signals—open auction listings from partner houses, localized market trend feeds, or lifestyle signals—always within consented frameworks.

Step 7: Institutionalize governance Create an internal review board to oversee AI outputs, ethical issues, and compliance. Schedule regular audits of models and training data.

This phased method limits risk while delivering incremental value.

Measuring impact: the KPIs that matter

Traditional vendor metrics emphasize throughput—messages sent or leads generated. For agent-centered AI, success looks different:

• Time per lead: measure how much less time agents spend on administrative tasks per lead.
• Re-engagement conversion: percentage of dormant contacts that respond within a targeted window after AI-assisted outreach.
• Conversion rate uplift: change in sales or listings attributable to AI-identified opportunities.
• Client satisfaction and trust: survey clients who received AI-personalized outreach for perceived authenticity and service quality.
• Agent adoption rate: percentage of agents using the tool weekly and editing AI outputs before sending.
• Revenue per agent: track whether AI leads to measurable revenue increases per agent.

A combination of efficiency and relationship metrics shows whether AI actually strengthens the human components of the business.

Common pitfalls and how to avoid them

Several risks repeat across deployments. Anticipating them reduces costs and reputational exposure.

Pitfall: Automating the agent’s voice

• Avoid building systems that attempt to perfectly emulate an agent’s language style without human oversight. Instead, offer editable drafts and clearly mark AI contributions.

Pitfall: Overreliance on public social data

• Public social signals are useful. They are incomplete and sometimes misleading. Combine them with direct client communications and explicit client preferences.

Pitfall: One-size-fits-all personalization

• Luxury clients expect nuanced engagement. Use granular taxonomies for interests (e.g., contemporary art vs. classical paintings) and avoid lumping diverse hobbies into broad buckets.

Pitfall: Ignoring compliance

• Don’t treat privacy laws as an afterthought. Embed consent workflows from day one and ensure data retention policies are enforced.

Pitfall: Measuring the wrong things

• If managers reward the volume of outreach rather than conversion or relationship health, agents will game the system. Tie incentives to meaningful outcomes.

Preventing these pitfalls requires a deliberate product roadmap and governance.

How smaller brokerages can get started without a full AI team

Wong noted that AI lowers the barrier to entry: thoughtful, intentional teams can use available tools equally well whether they have an in-house AI lab or not. Smaller brokerages should focus on practical, achievable steps:

• Use third-party CRM add-ons that provide prioritization and draft-generation features. Many vendors offer plug-and-play integrations that require minimal engineering.
• Create a simple rules-based personalization layer before committing to machine learning. For example, flag contacts with recent social activity or anniversaries and surface relevant templates.
• Outsource non-core infrastructure. Managed services for embeddings, vector search, or secure data storage speed deployment.
• Partner with niche providers. Luxury-focused services and auction feeds can be integrated via APIs to create curated signals without building the feed yourself.
• Train a pilot group of agents and iterate. Small, fast experiments highlight what works for your market and agents’ selling styles.

Even modest automation of administrative tasks can free enough agent bandwidth to produce measurable improvement.

The role of leadership: incentives, accountability, and long-term thinking

Leadership defines AI’s role. Companies that treat AI as a temporary efficiency play risk eroding trust. Leaders should commit to long-term integration focused on relationship value.

• Incentivize behavior that reinforces authenticity. Reward agents for high-quality conversations, not volume of AI-sent messages.
• Hold teams accountable to privacy and ethics KPIs, not just short-term revenue gains.
• Invest in both technical and human capital. Training budgets, ongoing model maintenance, and analyst oversight pay off.
• Communicate clearly to clients. Explain how personalization works and emphasize control and transparency.

Leadership alignment converts a promising pilot into an enterprise-grade advantage.

Where the industry is headed: agent roles, hybrid workflows, and the next decade

AI will not eliminate the agent role; it will reshape it. Expect these shifts over the next decade:

• Higher-value agent work. Agents will concentrate on negotiation, complex staging and pricing strategy, concierge services for luxury clients, and facilitating cross-domain purchases like art or collectibles.
• Specialized agent personas. Some agents will become luxury experience managers, leveraging data to curate art and lifestyle opportunities for clients between property transactions.
• Hybrid workflows. Agents will operate with an AI assistant that drafts, researches, and highlights opportunities but requires human finalization and relationship nuance.
• Greater data-driven personalization. The winners will be brokerages that synthesize property, lifestyle, and transactional data to anticipate client needs rather than react to them.
• New ethical norms. Industry standards will emerge for the acceptable use of client data and for agent representation when AI assists in outreach.

These trends favor firms that treat technology as an amplifier of human skill rather than a substitute.

Real-world signals: acquisitions, partnerships, and market context

Strategic moves reflect the logic behind Wong’s approach. Compass’s acquisition of Christie’s International Real Estate and other consolidation moves signal a market that values brand synergies and data integration. Some brokerages that experimented with early conversational automation found the approach inadequate and are pivoting to augmentation. Those market signals suggest that the path to sustainable advantage lies in connecting complementary assets—luxury auction houses, curated lifestyle data, and high-touch sales—through technology.

That integration creates new business models: agents maintain contact by surfacing relevant lifestyle opportunities, auction houses access high-net-worth buyers with real estate ties, and brokerages differentiate through curated client experiences. Firms that can operationalize these connections on a consent-forward, secure platform will capture outsized loyalty.

Practical example: what an AI-assisted luxury touchpoint looks like

This illustrative workflow shows a realistic agent-client interaction enabled by AI without attributing specifics to any company beyond Wong’s publicly stated approach:

1. Data ingestion: The system imports the client’s CRM record, email history (with consent), and public social posts.
2. Signal detection: The algorithm identifies interest in contemporary art and a recent comment about a European trip.
3. Auction feed match: Christie’s auction calendar shows a lot by an artist the client follows in two weeks.
4. Agent prompt: The agent receives an alert: “Client X follows Artist Y; Artist Y’s work is on auction next week. Suggested outreach: congratulate on recent trip, mention the auction, and offer to provide a preview link.”
5. Agent edits and sends: The agent customizes the suggested message to match voice and sends it. The message feels authentic because it references a personal detail the client had previously shared.
6. Outcome tracking: The system logs the interaction and tracks whether the client engaged with the auction or re-opened conversation about property needs.

This scenario keeps the agent in control and uses AI to deliver timely, relevant reasons to connect.

Building trust with clients: messaging and consent strategies

Personalization at scale depends on trust. Clear, simple communication about how data will be used prevents misunderstanding.

Best-practice messaging examples

• “We use your preferences to recommend art and lifestyle opportunities that match your interests. You can opt out at any time.”
• “Would you like us to notify you about auctions related to artists you follow? This helps us share timely items you may enjoy.”
• “We’ll only use information you’ve shared with us or posted publicly; private messages are treated confidentially and require your consent to analyze.”

Make consent granular: allow clients to pick which categories they want personalization for (real estate only, art and auctions, lifestyle events, etc.). Provide an easy path to review and delete personal data. These steps preserve client autonomy and strengthen long-term relationships.

The economics: where the value accrues

AI delivers three economic advantages when properly aligned with agent workflows:

• Efficiency gains. Automating repetitive work reduces agent hours spent on admin and raises the number of client-facing interactions per week.
• Higher conversion. Personalized outreach increases response rates and the likelihood of re-engagement among dormant contacts.
• Cross-sell opportunities. Connecting auction feeds and luxury interests with real estate relationships unlocks revenue beyond the transaction—introductions to auctions, concierge services, or curated experiences.

Leaders should track both direct transaction revenue and indirect income from cross-domain engagement to capture the full economic picture.

What to ask a vendor or partner: procurement checklist

When evaluating AI vendors or partners, prioritize features that reflect agent-centric design:

• Does the system require agent approval for all client-facing content?
• How does it collect and manage consent? Is it granular?
• What signals does it ingest, and how accurate is identity resolution across platforms?
• How explainable are the propensity scores and recommendations?
• Is there an audit trail for generated content and edits?
• What security certifications and data residency options are available?
• Does the vendor offer ongoing model monitoring and bias audits?
• What training and onboarding support is included?

Procurement decisions should balance ease of deployment, alignment with ethical practices, and measurable impact on agent productivity.

The competitive edge: why authenticity beats mimicry

Impersonation may seem like a shortcut to scale but it undermines the single strongest asset a broker has: agent authenticity. Agents who preserve their voice and leverage AI to deepen client knowledge create durable competitive advantages. The real opportunity lies in amplifying the traits that win referrals—trust, empathy, and context-specific advice—rather than manufacturing a synthetic substitute.

Wong’s experience demonstrates that a careful, human-centered approach drives both adoption and client satisfaction. That lesson applies broadly: building systems that respect identity, consent, and the nuances of luxury relationships generates more value than cold automation ever could.

Looking ahead: practical signals to watch for industry-wide change

Several indicators will reveal whether the industry is moving toward authentic augmentation or toward risky automation:

• Regulatory action on synthetic impersonation and disclosure in business communications.
• Widespread adoption of consent-first personalization flows in brokerages and auction houses.
• Growth of cross-domain data partnerships—luxury retailers, auction houses, and brokerages—combined with secure data-sharing frameworks.
• Emergence of industry standards for auditing AI models used in client-facing scenarios.
• A rise in agent specialization toward curated lifestyle management and concierge services.

Watching these signals helps leaders prioritize investments that preserve client trust while improving productivity.

FAQ

Q: Can AI really replace agents in real estate? A: No. AI cannot replace the nuanced, trust-based work agents perform in negotiations, local market advising, emotional support, and building long-term relationships. The most effective systems free agents from repetitive tasks so they can spend time on these high-value activities.

Q: How does @properties Christie’s International Real Estate use AI to reconnect past clients? A: They ingest historical emails and public social media signals to surface personalized conversation prompts and prioritize outreach. The system suggests why to contact a client and provides editable message drafts, enabling agents to re-engage with contextually relevant content.

Q: Is it legal to use social media posts for personalization? A: Public social posts are generally permissible to use, but privacy laws and platform terms vary. Best practice is to be transparent about data use, obtain explicit consent for analyzing private communications, and provide easy opt-outs.

Q: How do you prevent AI from impersonating an agent? A: Design the system so agents review and approve every client-facing message. Avoid automatic voice cloning or unsupervised messaging, maintain editable templates, and clearly disclose when content is AI-suggested.

Q: What are measurable benefits brokerages can expect? A: Expect time savings on administrative work, higher re-engagement rates among dormant contacts, improved conversion from AI-assisted outreach, and potential cross-sell revenue from curated luxury interactions.

Q: What should small brokerages do first? A: Start with plug-and-play CRM enhancements that prioritize outreach and automate research. Run a pilot with a small group of agents, focus on transparency and consent, and scale features that demonstrate measurable time savings and conversion lifts.

Q: How should brokerages measure AI success? A: Combine efficiency metrics (time saved), relationship metrics (re-engagement and conversion rates), and financial metrics (revenue per agent). Also track client trust indicators through surveys and monitor agent adoption rates.

Q: What governance is necessary? A: Establish an internal review process for AI outputs, enforce consent and opt-out mechanisms, schedule regular audits of model fairness, and ensure secure handling and encryption of client data.

Q: Will personalization at scale erode the luxury brand? A: If done poorly, yes. If executed with consent, discretion, and agent control, personalization enhances the luxury brand by providing curated, relevant experiences that demonstrate real knowledge of client interests.

Q: How will agent roles change with AI? A: Agents will focus increasingly on high-touch, strategic roles: negotiation, local expertise, concierge services, and curation of lifestyle opportunities. AI will handle background tasks and surface timely reasons to connect.

Q: What are the top risks and how to mitigate them? A: Risks include privacy violations, impersonation, biased models, and over-automation. Mitigate through explicit consent, human-in-the-loop workflows, regular bias audits, and governance that prioritizes relationship value over automation volume.

Q: How can brokerages integrate auction house data like Christie’s? A: Create consent-based feeds that link client interest tags to auction calendars. Trigger agent notifications when matched items appear, and provide editable message templates to facilitate authentic outreach.

Q: What are realistic timelines for rollout? A: Small pilots can show results in weeks for basic automation; robust personalization and cross-domain integration typically require several months to a year, depending on data maturity and governance processes.

Q: How do you guard against biased propensity models? A: Monitor outcomes across demographic groups, use explainable features, avoid unnecessary demographic inputs, and retrain models with fairness objectives and human oversight.

Q: What distinguishes meaningful personalization from intrusive surveillance? A: User control and consent are decisive. Meaningful personalization amplifies client preferences the client has explicitly shared or consented to analyze. Intrusive surveillance relies on covert aggregation and use of data without clear client awareness.

Q: What should clients ask their agents about AI use? A: Clients should ask whether their communications or social data will be used to personalize outreach, how to opt out, and how the brokerage protects and deletes personal data. Agents who answer clearly demonstrate respect for client autonomy.

Q: Will these AI tools create new revenue streams beyond commissions? A: Yes. Curated introductions to auctions, concierge services for art and collectibles, and lifestyle experiences can generate fee-based revenue and deepen client relationships between property transactions.

Q: What is the single most important lesson from Wong’s experience? A: Preserve the human connection. AI should remove friction and enhance agent insight, not pretend to be the agent. Authenticity wins.

---

This account of how @properties Christie’s International Real Estate shifted its AI strategy shows a path forward: protect what makes agents indispensable and apply technology to expand their capacity. The result is not a threat to the agent role, but a redefinition of how value is delivered—more precise, more timely, and more aligned with each client’s interests.