2025 · Real estate · Lead · Leon & Vera
Leon & Vera — an AI lead team for real-estate brokers
AI specialists working over WhatsApp for European real-estate brokers: scouting owner-sold listings, qualifying leads, running valuation campaigns and managing the ads behind them.
Live in 6 European markets with subscription billing, GDPR and EU AI Act compliance
Context
A broker wins a listing by being the first credible contact the seller speaks to, and the sellers worth reaching — the ones selling privately, without an agent — are scattered across listing sites and local classifieds. Finding them, contacting them and qualifying them is manual work that competes with the rest of a broker's day, so it happens late, inconsistently, or not at all.
Doing it automatically across Europe brings its own constraint: automated contact and AI disclosure are regulated, the rules are not identical from one market to the next, and getting them wrong is not a bug report but a fine.
Decisions
Chose WhatsApp as the channel, inside the Business API's template and opt-in rules over email sequences and web forms
A private seller answers WhatsApp and ignores an email from an agency they have never heard of — and when the listing goes to whoever makes credible contact first, reply rate is the whole product. The cost is working inside WhatsApp's template approval and opt-in constraints, which shapes what the system is allowed to say and when.
Chose A roster of specialists — scouting, qualification, valuation, ads — each with its own remit over a single general assistant handling the whole funnel
The jobs fail differently. Scouting is a retrieval and matching problem, qualification is a conversation with a person who can be annoyed, and ad management spends money. Separating them means each can be constrained, measured and corrected on its own terms instead of one prompt being tuned against four conflicting objectives.
Chose Compliance designed into the data model — disclosure, consent, retention, record-keeping over shipping the product and adding a compliance layer before launch
GDPR and the EU AI Act decide what may be stored, for how long, and what a recipient must be told before the conversation starts. Those are properties of the schema and the message flow, not a feature to be added on top; retrofitting them means migrating conversations that were recorded under rules they did not follow.
Chose Subscription billing per broker over taking a commission on each lead or closed listing
A commission model makes the platform a competitor for the relationship the broker is trying to own, and it invites arguments about attribution on every deal. A flat subscription keeps the incentive simple: the broker keeps every lead, and the product has to be worth its fee on volume alone.
What I did
I built the platform as a team of specialists rather than one assistant: scouting owner-sold listings, qualifying the leads that come back, running valuation campaigns, and managing the advertising behind them — each with its own job and its own guardrails.
Conversation runs over the WhatsApp Business API, inside its template and opt-in rules, with the assistant disclosed as AI. Around that sit subscription billing, per-broker data handling built for GDPR, and the disclosure and record-keeping obligations of the EU AI Act — designed in rather than added before launch, because the product went live in six markets at once.
Outcome
Live in six European markets, with brokers running lead generation as a subscription rather than a per-lead commission — the leads, and the relationships, stay theirs.
In hindsight
The compliance surface was treated as one thing and it is not: disclosure wording, consent, retention and record-keeping all vary by market, and I built them as branches in code before recognising they are configuration. Modelling market rules as data from the start would have made the sixth market a settings screen instead of a release. The general lesson is one I now apply early: anything that varies per jurisdiction is data, not logic.