Prepared for Rock — working US realtor evaluating where to build.
The realtor-AI market in April 2026 is loud, crowded at the top, and shockingly thin in the middle. Dominant operators: Compass ($1.8B over a decade on platform; merging with Anywhere) (Compass), BoldTrail (rebranded kvCORE; brokerage/team tooling at $499–$1,800/mo), Lofty (just launched the first "agentic AI operating system" Feb 2026) (RISMedia), and Follow Up Boss ($58–$833/mo, native AI bundled). Bottom tier: Wise Agent ($32/mo), IXACT ($47/mo), the ailing Top Producer. What's missing is a thin layer aimed at the individual realtor doing buyer-side work — almost every "AI" product is a lead-funnel chatbot or brokerage-grade ops platform. The gap is the buyer-agent's middle game: calls-to-listing-agents, comp pulls, price-defense narratives, inspection-followup. No major player ships there.
End-to-end Compass-only agent platform — CRM, marketing, CMA, voice-activated assistant that drafts emails, builds CMAs, generates property presentations, and surfaces "Buyer Demand" signals (Compass). Pricing: bundled into Compass's brokerage split. Founded: 2012; NYSE:COMP; $1.8B platform spend over 10 yrs (HousingWire). Weakness: Locked to Compass agents. CEO Reffkin's MLS antagonism is creating friction (Real Estate News). Useless outside the walled garden.
All-in-one CRM + lead gen + marketing. Now ships Lofty AOS, an agentic system with named sub-agents (Sales, Social, Homeowner) that autonomously execute workflows (Lofty AOS). Pricing: $449/mo core scaling to $1,500/mo (Capsule CRM). Founded: 2016 as Chime; rebranded 2023. Weakness: "Some users find it pricey, especially for solo agents, with lack of transparent pricing... and occasional bugs with support that can be slow to respond" (TheClose).
AI lead-qualification chatbot ("Aisa Holmes") texts and calls inbound leads for up to 12 months. Pricing: Explore $499/mo (200 leads), Elevate $999, Established $1,499, plus setup + carrier fees (Unify). Founded: 2016, Ames IA. Weakness: Too expensive for solos; only worth it at high lead volume. Competitors (Ylopo's Raiya, Lofty's Sales Agent) now bundle the same feature.
The agent-loved CRM. Clean inbox, fast routing, deep integrations. Pricing: Grow $58/mo, Pro $416, Platform $833 (FUB). AI: native, no extra charge — smart summaries, smart messages, predictive prioritization. AI-Calling is a paid add-on. Founded: 2011; acquired by Zillow Group 2023. Weakness: Not a lead-gen system — bring your own leads. Per-user pricing on Pro/Platform stings teams. Some agents resent Zillow ownership.
All-in-one CRM, IDX, lead gen, marketing automation, primarily for teams/brokerages. Pricing: quote-based, ~$499–$1,800/mo plus $500–$1,000 setup (JustARealtor). Founded: 2008; rebranded BoldTrail mid-2024. Weakness: Brutal reviewer quotes. "The CMS is terrible and buggy." "We entered the contract based on promises... but the system has been unstable, glitchy, and poorly supported from the start." "DO NOT pay for their Lead 350 program as it bogs down your email and text messages with bogus leads." (G2)
CRM + IDX + lead gen with AI lead engagement. Pricing: solo $299.95/mo; team $359.95–$724.95 (Sierra). Founded: 2007. Weakness: Solid mid-tier but doesn't lead any category — feels like a backup choice.
AI-powered ad platform + IDX + lead nurture (not a CRM). Pricing: ~$300–$1,000/mo for small teams; 6–12 month contracts (inboundREM). AI: Raiya texting, AI Voice (claims 45% answer rate over 14 attempts), dynamic FB ads. Founded: 2016. Weakness: Needs a CRM alongside. Long contract lock-ins.
CRM + lead gen + IDX, premium-tier for established teams. Pricing: $1,000–$1,700/mo + $1,500 setup + $1,000/mo mandatory ad budget — all-in ~$2,500/mo floor (Capterra). Founded: 2006. Weakness: "Dissatisfied customers typically reference billing issues, like baiting-and-switching, overcharging... and poor quality leads with lots of fake phone numbers."
IDX websites + CRM + lead gen for solos and small teams. Pricing: $299/mo for two users, $25 per added agent. Geek AI bundled into Grow tier late 2025 (Real Geeks). Founded: 2007. Weakness: Cookie-cutter sites. $798 two-month buyout fee documented in BBB complaints. "Cancel and you lose access to the website and all content hosted there."
Buyer-facing listing alerts + collaborative search + market activity alerts. CRM-agnostic — sits on top of your CRM. Pricing: quote-based; ~$60–$120/mo per agent per Capterra reviewers. Founded: 2012. Weakness: Does one thing well (database nurture). Useless without an existing database.
Budget CRM with marketing automation, transaction templates, recent AI Writing Assistant. Pricing: $32/mo monthly, $27.17/mo annual (Wise Agent). Founded: 2002. Weakness: Dated UI; AI feels bolt-on. Loyalty held by best-in-class support.
DISCONTINUED Sept 2025. Owner Lone Wolf Technologies sunset it and migrated users to "Lone Wolf Relationships" (RealtyJuggler). Historical $25–$99/mo. A long-running CRM with thousands of solo-agent users killed by its parent — the cautionary tale of the year.
Oldest real estate CRM (1982). Pivoted to social-listening seller-prospecting via "Smart Targeting." Pricing: from $40/mo. Weakness: UI looks like 2014. Younger agents skip it.
ListedKit AI (Ava) — AI transaction coordinator. Drag in any state's purchase agreement (even handwritten); Ava extracts dates, builds timeline, syncs Calendar, drafts emails. $9.99/transaction, no monthly fee, first one free. 2025 Inman Innovators Award Finalist; 2,100+ transactions processed (ListedKit). Most interesting newcomer — completely orthogonal to the CRM-stack mindset.
HomeLight EVA — AI agent automating closings end-to-end. Raised $40M in 2025 (Yahoo). Closing-coordinator workflow, not agent-facing.
Lofty AOS (Feb 2026) — first major incumbent shipping truly agentic AI. Industry significance > current usefulness.
Rexera — AI agents for closing workflows (HOA docs, lien searches). Title/escrow flavored.
Legend: ✅ = solid native capability · ⚠️ = partial, weak, or paid add-on · ❌ = absent
| Tool | Lead chat AI | Drip nurture | Buyer listing alerts | Transaction mgmt | Comps / valuation | Appt scheduling | IDX integration | Mobile app | Open API |
|---|---|---|---|---|---|---|---|---|---|
| Compass AI | ⚠️ | ✅ | ✅ | ⚠️ | ✅ (CMA gen) | ✅ | ✅ | ✅ | ❌ (closed) |
| Lofty (AOS) | ✅ | ✅ | ✅ | ⚠️ | ⚠️ | ✅ | ✅ | ✅ | ⚠️ |
| Structurely | ✅ | ✅ | ❌ | ❌ | ❌ | ✅ | ❌ | ⚠️ | ✅ |
| Follow Up Boss | ✅ | ✅ | ❌ | ⚠️ | ❌ | ✅ | ⚠️ | ✅ | ✅ |
| BoldTrail (kvCORE) | ✅ | ✅ | ✅ | ⚠️ | ⚠️ | ✅ | ✅ | ✅ | ⚠️ |
| Sierra Interactive | ✅ | ✅ | ✅ | ❌ | ❌ | ✅ | ✅ | ✅ | ⚠️ |
| Ylopo | ✅ (Raiya) | ✅ | ✅ | ❌ | ❌ | ⚠️ | ✅ | ✅ | ⚠️ |
| BoomTown | ⚠️ | ✅ | ✅ | ❌ | ❌ | ✅ | ✅ | ✅ | ⚠️ |
| Real Geeks | ✅ (Geek AI) | ✅ | ✅ | ❌ | ❌ | ⚠️ | ✅ | ✅ | ⚠️ |
| RealScout | ❌ | ⚠️ | ✅ (best-in-class) | ❌ | ⚠️ | ❌ | ⚠️ (overlay) | ✅ | ✅ |
| Wise Agent | ⚠️ | ✅ | ⚠️ | ✅ | ❌ | ✅ | ⚠️ | ✅ | ✅ |
| Top Producer | ⚠️ | ✅ | ⚠️ | ✅ | ❌ | ✅ | ⚠️ | ✅ | ⚠️ |
| ListedKit AI | ❌ | ❌ | ❌ | ✅ (best-in-class) | ❌ | ⚠️ | ❌ | ⚠️ | ⚠️ |
| HomeLight EVA | ❌ | ❌ | ❌ | ✅ (closing only) | ❌ | ❌ | ❌ | ❌ | ⚠️ |
The empty columns: Note how few tools touch comps/valuation honestly (Compass is the only "solid"), and how many show ⚠️/❌ on transaction management. Almost nothing handles buyer-side intelligence (listing-agent calls, mechanical-info extraction, price-defense narratives).
The market splits into four bands:
AI surcharges are disappearing. Real Geeks moved Geek AI from a $200/mo add-on into the Grow tier. FUB never charged extra. Lofty AOS is included for paying customers. AI is table-stakes now, not premium. The "we'll charge $200/mo for an AI add-on" positioning is dead. Sole exception: AI calling (Ylopo, Structurely voice, FUB Calling) still costs extra because of Twilio carrier + compliance fees. That's the one durable "AI tax."
Synthesizing the reviewer evidence:
1. Lead-quality fraud and disappointment. The single most repeated complaint across BoomTown, BoldTrail, Real Geeks: paid-lead programs deliver garbage. "DO NOT pay for their Lead 350 program as it bogs down your email and text messages with bogus leads." (BoldTrail reviewer). "Poor quality leads with lots of fake phone numbers." (BoomTown reviewer).
2. Customer-support collapse at scale. "The support is horrendous, they make up answers, a solution that should take 10 minutes is an all day event." (BoldTrail). "Wait times can be up to an hour and then they tell you a specialist will call you." These are paying customers at $500–$1,800/mo.
3. Hostile cancellation terms. Real Geeks charges a documented $798 two-month buyout. You also lose your website and all hosted content on cancellation. Ylopo locks you into 6–12 month contracts. The industry has a "land and trap" pattern.
4. Glitchy software at premium prices. "The CMS is terrible and buggy, and the platform suffers from glitches and slow loading and processing times." (BoldTrail). At $1,500+/mo, agents expect the basics to work.
5. AI that's mostly chatbot, not assistant. Outside Lofty AOS and Compass AI, the "AI" in most products is a single-purpose lead-text-bot. Realtors who tried these tools report them as glorified auto-responders that don't understand context, don't know the local market, don't follow up well past the first 2–3 messages.
6. Photo manipulation embarrassments. Late-2025 industry articles flag agents posting AI-edited listing photos that misrepresent properties — sunny rooms that are actually dim, and one viral case of a "demonic figure emerging from a mirror" in a generated bathroom photo (Futurism). The whole space has a credibility hangover.
7. Solo agents are second-class citizens. Pricing tiers, feature design, and onboarding flows are built for teams. Solo agents pay disproportionately or get the stripped version.
8. Failed entrants haunt the space. LionDesk's September 2025 shutdown stranded thousands of paying solo agents. The lesson: even a 15-year incumbent at $25–$99/mo can be killed by its parent overnight.
Five specific underserved angles, ranked by how few competitors touch them:
Every buyer-agent does this manually: text or call the listing agent, ask "how old is the roof, when was the AC replaced, any recent updates, what does the seller want, are there other offers." Nothing on the market automates this. Compass's voice assistant lets you draft outreach; it does not make the call and extract structured answers. With voice-AI now functional (Bland, Vapi, Retell), this is an obvious wedge — and zero of the 14 competitors ship it. The closest adjacency is HomeLight EVA, but that's closing-side, not buyer-side discovery.
MLS-pulled comps are still done manually because tools either (a) can't legally access MLS data, or (b) charge brokerage-wide deals (Compass, BoldTrail, BoomTown) that lock data inside their walled garden. Agents say "agents want better market and consumer insights, AI pricing tools and more" from MLSs that aren't shipping them (Real Estate News). A tool that authenticates as the agent into their MLS and produces a defensible comp report in 90 seconds would have enormous pull.
Every AI tool generates listing copy for sellers. Zero generate the buyer-agent's counterargument: "Here's why this house is overpriced based on these 6 comparable closings, the days-on-market trend in this submarket, the inspection issues we found, and the comparable price-per-square-foot." Buyer agents write this from scratch every time. The market is asymmetric — listing agents have abundant tooling, buyer agents have nothing built for their negotiation workflow.
Per-transaction TC fees are $250–$600/file; full-time TCs cost $300–$500/mo retainer (Paperless Pipeline). ListedKit AI (Ava) at $9.99/transaction is the only product priced for the solo agent doing 8–20 deals/year. There's room for a competitor that pairs Ava-style contract intake with proactive deadline chasing (texting the lender, the inspector, the title company at the right moments).
The big platforms are built by software founders who hired realtor advisors. Almost every reviewer complaint reduces to "they don't get how I actually work." A product built first by a working realtor, for themselves, then extended to friends, has a credibility moat the funded incumbents cannot easily attack. This is the wedge BoomTown, FUB, and Compass cannot pursue — they already hold the enterprise sales motion they built their company around.
Given the gaps, here's what a new entrant would need to do:
Starting wedge — pick one workflow, not a platform. The fatal mistake is launching "an AI for realtors." Every funeral in this space (LionDesk being the freshest) was a generalist that got squeezed. Pick the single most painful workflow for a specific sub-segment. The best candidates from the gaps above, ranked:
1. Listing-agent voice-AI for buyer-side discovery. Build a tool that, given an MLS link, places a call to the listing agent's phone, runs a 4-minute scripted conversation extracting roof age, mechanical history, recent updates, seller motivation, and competing-offer status, and returns a structured report. Charge $15–$30/listing or $99/mo for unlimited. This is a knife, not a platform.
2. Buyer price-defense generator. Pull comps via MLS auth, write the counter-narrative, package as a PDF for the buyer client. $49/mo.
3. Solo-agent TC competitor. Compete with ListedKit AI on workflow depth (post-contract chase) rather than price.
Distribution — the warm-network wedge. Forget paid ads. The successful real-estate-tech companies (FUB especially) grew through agent referrals and live demos at brokerage events. Path:
Pricing strategy — undercut the AI-add-on tier. $39–$99/mo lands below FUB Pro ($416), Lofty ($449), Sierra ($299), but above Wise Agent ($32). Realtors expense everything; the resistance point is at the brokerage justification level, not the dollar amount. Price for the agent's monthly mental budget, not for what comparable feature-stacks charge.
Defensibility — workflow integration, not data. In real estate, data moats are weak (MLS data is licensed, public records are public, inspection reports are private). Brand moats are weak (agents switch tools every 18 months). The only durable moat is becoming the muscle memory of the daily workflow — the one tool the agent opens when a new listing hits their MLS feed. Speed-to-result and integration depth (Calendar + Phone + CRM + DocuSign + e-mail) are what creates lock-in. A good rule of thumb: if your tool can be replaced in 30 minutes, it's not a moat.
What kills startups in this space:
1. The "AI for realtors" generalist trap. Most-common failure pattern. The graveyard includes a long tail of mid-decade chatbot startups. Founders pitch "AI for the entire workflow," try to build CRM + lead gen + nurture + transaction, and run out of money before any one feature is best-in-class. Lofty AOS just took the entire generalist territory in February 2026 — that game is over for newcomers.
2. The iBuying / valuation collapse. Zillow Offers was the cautionary tale; Casavo collapsed in Europe. AVM-driven business models keep failing because the market punishes pricing errors brutally (PropTech in Europe). Don't build "Zillow but with better AI."
3. The MLS-data lawsuit. Multiple proptech startups have been sued for scraping MLS data. Any tool that bypasses MLS authentication is one cease-and-desist away from death. Get the agent to log in themselves; never store or re-distribute MLS records.
4. The agent commission turmoil. Post-Burnett-settlement (the 2024 NAR commission case), agent income models are unstable. This is good for tools that save time (agents are cost-conscious) and bad for tools that ride on commission-based pricing.
5. The proptech funding winter. PropTech funding is down >50% from its 2021 peak (Crunchbase). Don't build a story that requires Series A in 12 months.
6. The discontinuation risk. LionDesk's September 2025 shutdown by Lone Wolf reminds us: even revenue-positive incumbents with thousands of paying users can be killed for portfolio reasons. As a founder, this is a moat (incumbents are skittish about new investment); as a customer, it's a reason to bias toward indie tools that won't be portfolio-pruned.
The 2–3 successful patterns instead:
For a working solo realtor with a 13-year assistant relationship, deep market knowledge, and willingness to be the first user — three concrete options:
Option A — Build the listing-agent voice-AI (Gap 1).
Option B — Build the buyer-side price-defense generator (Gap 3).
Option C — White-label your assistant relationship.
It's the cleanest test of whether "build it for myself first" creates a sellable product. The workflow is narrow (one call, structured output), painful (every buyer-agent does this 5–20×/month manually), demoable (live call in 4 minutes at any board lunch), defensible (the script is the moat — you'll iterate across hundreds of real calls before any competitor does), and priced for the solo agent (sub-$30/use, no contract). Voice-AI infrastructure is cheap enough to ship v0 in 2 weeks.
It dodges every failure pattern in §8: not a generalist, not an AVM, doesn't scrape MLS, doesn't need Series A, isn't a portfolio-pruning target because it's yours. If it scales past 50 paying agents — real business. If not — you've still built yourself 10 hrs/week back, which covers the build cost in a month.
Sources cited inline. Where review fragments appear, they were drawn from G2, Capterra, BBB, and TheClose syntheses — verifiable but not always with a single canonical URL because most sites aggregate dozens of reviews.