How High-Performing Real Estate Teams Are Actually Using AI (Straight From 4 Top Producers)

AI in real estate has been the topic of conversation for a while now, but there’s a real gap between talking about AI and actually running a business on it. A panel of four top producers, Adam Gillespie (Apex Elite AI, Inman’s first AI Award winner), Katie Van Ness (Most Referred Expert community, Lofty Masterclass), Zach Geisendorfer (Fam Realty Group, San Diego/Orange County), and Tristan Almada (Why Realty), sat down for a candid conversation on what’s actually working right now, not theory, but the specific systems running inside their businesses today.

The Shift: From “Nice to Have” to “How I Run 95% of My Business”

Adam’s story sets the tone for where this is all headed. Before AI, he built every follow-up campaign by hand, spending weeks writing a single five-year post-closing drip. After discovering ChatGPT in late 2022, he went all-in on formal training in AI ethics, prompt engineering, and large language models, then rebuilt his business around it.

Today, roughly 95% of his business runs on AI, with him acting as “human in the loop.” He operates like he has a team of 10-15 people behind him, when in reality it’s him, a small set of AI agents, and one transaction coordinator who does very little manual work.

The bigger industry prediction: single agents who go all-in on AI are going to start posting multi-million dollar volume years with minimal staff, reshaping what “solo agent” even means over the next few years.

Follow-Up: Where Most Agents Are Actually Failing

Adam was direct about this: follow-up, not lead generation, is where 99% of agents fail. Agents blame “bad internet leads” when the real issue is a missing follow-up system.

His five-ingredient follow-up framework for every lead:

1. Automated property alerts the moment a lead enters the CRM
2. A 10-day text campaign attempting to set an appointment (pauses automatically the moment someone replies, so a human takes over)
3. A two-year informational email drip, written in the agent’s own voice and tone (cloned into AI), that builds trust rather than pushing urgency
4. Personalized holiday cards (including AI-generated, genuinely funny custom images) that reinforce the agent is an advisor, not just a salesperson
5. An automated call schedule running 603 days across 36 total call attempts, because data shows internet leads often don’t answer until the seventh call, and most agents give up after two or three

One real result from this system: a deal closed on a six-year-old lead who had never once answered a call, text, or email, until an inbound showing request came in out of nowhere.

The underlying insight on trust: realtors rank as the third least-trusted profession, behind attorneys and used car salesmen. People want to buy homes, they just don’t trust the agent yet. Long-form, consistent, low-pressure content over time is what closes that gap.

Using AI Without Losing the Human Relationship

Katie raised an important caution: some agents automate their “sales seat” too early, before someone (even themselves) is actually staffing that seat, and their businesses quietly collapse a few months later. AI should remove mundane, repetitive work, not replace genuine connection at the exact moments that require it.

Her bigger shift was going deeper with AI than most agents do: instead of only asking for a listing description or social post, she now feeds AI the full picture, revenue sources, goals, unique value proposition, and overall business strategy, consolidated into one ongoing orchestrator conversation rather than siloed, one-off prompts.

She also flagged something worth taking seriously: constant AI interaction can create a low-grade “dopamine bleed” that contributes to burnout. Her fix is finishing tasks completely (even small ones) rather than spreading attention across many half-finished threads.

Real Client Communication at Scale: Storytelling Over Stats

Zach’s team runs roughly 1.9 million contacts through automated “smart plans” in Lofty, each triggered by a different data point (FSBOs, expireds, absentee owners, notice-of-sale, and more), each following a distinct journey and sales process.

The key differentiator isn’t volume, it’s storytelling. Rather than broadcasting “just sold” posts, his team turns every win into a narrative: the buyer who got 11 offers in a week, the seller who accepted a cash offer and had funds in 10 days. AI is used specifically to turn a bare fact into a story format, because people picture themselves inside a story in a way they simply don’t with a stat. That same storytelling extends to listing landing pages: updating a “just sold” page with the actual story behind the sale (open house attendance, offer count, final sale price relative to asking) rather than a flat announcement.

AEO Is the New SEO, and It’s Working Right Now

Tristan shared live search results showing hyper-local blog content (like “best kid parks in Conejo Valley” and a Father’s Day roundup) landing directly in Google’s AI Overview and on page one, despite being posted only weeks earlier. One article alone generated over 10,000 views.

What’s actually driving this:

• Consistent identity across the web. The same name, domain, and local market association everywhere strengthens authority.
• Proper schema (JSON-LD), which can be automated through Google Tag Manager so every new blog post generates correct schema without manual work.
• Answering the question immediately. AI models scanning content are, in Adam’s words, “efficient but lazy.” If the question isn’t answered in the first paragraph, the model moves on regardless of how good the rest of the content or schema is.
• GEO (Generative Engine Optimization): backing up the answer with specific facts and data in the middle of the post, with Q&A formatting toward the end.

A practical content-generation approach: use AI to research the 50 most commonly asked questions for a specific buyer or seller persona in a specific local market, then turn each question into multiple content formats, a blog post, an Instagram carousel, short-form video, and long-form YouTube content, stacking several pieces of indexable content from a single research pass.

One more platform-specific note: indexing quota matters. If a website’s IDX listings consume the available Google indexing quota, blog content may never get indexed at all. Prioritizing blog and custom landing pages over raw listing pages (which shouldn’t be indexed anyway, since they go stale) makes a meaningful difference.

Tracking the Metric Most Teams Ignore: Speed

Zach’s team tracks a metric most teams overlook entirely: speed, specifically, how quickly a new agent or a new lead moves through the pipeline. Their two-week onboarding boot camp (compared to a typical 90+ day ramp) now produces an 88% success rate for new agents signing their first agreement within two weeks, largely because AI-powered knowledge bases answer common onboarding questions instantly instead of requiring a wait for a sales manager.

The broader principle: any place AI can compress time between a lead entering the system and a real conversation happening, or between hiring an agent and that agent generating GCI, is a direct lever on revenue.

Try Lofty for Yourself

Every AI feature discussed throughout this panel, sales assistant automation, smart plans, blog schema automation, is built into Lofty. There’s a special partner link for this community:

👉 Get 10% Off Lofty

Key Takeaways

• Follow-up, not lead generation, is where most agents actually fail. A structured, multi-channel, long-duration follow-up system (calls, texts, email drips, even holiday cards) captures deals most agents write off as dead.
• Internet leads often don’t answer until the 7th call. Giving up after 2-3 attempts abandons money already spent on lead generation.
• AI should replace mundane, repetitive tasks, not the sales relationship itself. Automating the “sales seat” too early can quietly stall a business.
• Storytelling outperforms stat-dropping in nurture content, and AI can turn a plain fact into a narrative format instantly.
• AEO (Answer Engine Optimization) is functioning like early-days SEO. Hyper-local, question-first content with proper schema is landing in AI Overviews within weeks.
• Answer the question immediately in your content. AI models skip content that buries the answer, no matter how strong the schema is.
• Track speed as a core business metric. Faster lead response and faster agent onboarding both convert directly into revenue.
• Be transparent about AI usage. NAR has signaled that brokerage AI policies are coming by the end of 2026, and agents remain responsible for how AI is used, not the AI provider.

Next Steps

Pick one thing from this list to implement this week: either build out a real multi-touch follow-up sequence for your next lead source, or write one hyper-local blog post answering a specific question your ideal client is actually searching for, then check where it lands in a week.

• Watch the full YouTube Webinar Replay here
• Get 10% Off Lofty