I Let My AI Assistant Hunt Apartments While I Made Lunch — Here’s What It Found

Monday, March 23, 2026 — Tel Aviv

Five minutes. That’s how long it took BruBot to find five apartments in Kochav HaTzafon — the north Tel Aviv neighborhood I’ve been stalking for months — while I did literally nothing but exist. No browser tabs. No scrolling Facebook groups at 11 PM. No “let me just check one more time” spirals.

Just: “find me apartments,” and then results.

Here’s what the hunt looked like.

The Old Way vs. The New Way

For months, apartment hunting in Tel Aviv has meant:

Bookmarking 14 different Facebook groups
Filtering by neighborhood, price, rooms, and “mamad” (bomb shelter) — manually, every single day
Getting emotionally invested in a listing at 11 PM, discovering it’s gone by 9 AM
Losing two hours you didn’t have to spend

It’s not a search problem. It’s a signal-to-noise problem. Kochav HaTzafon has maybe 10 genuinely good listings at any given time, buried inside hundreds of irrelevant posts, duplicates, and listings that disappeared three days ago.

What BruBot Does Differently

Three layers of filtering, stacked intelligently:

Layer 1 — Raw Scraping: Hits Facebook Marketplace, Yad2, and dedicated rental aggregators simultaneously. Gets everything. No filter, no judgment. Just data.

Layer 2 — Smart Pre-Filter: Removes the obvious noise — out-of-budget, wrong neighborhood, studio apartments when you need 3BR, listings with prices that make no sense. This kills roughly 70% of everything immediately.

Layer 3 — GPT Evaluation: What remains (usually 10-30 listings) gets evaluated against your actual criteria: neighborhood, size, price per sqm, whether the description sounds legitimate or like a scam, proximity to what matters to you. Each listing gets scored 0-10.

Only the 8+/10 listings make it to your phone. Everything else never even wakes you up.

Today’s Results

Five listings came back. Here’s the shortlist:

90sqm on Daphne Street — ₪7,800/mo, North TLV. Best value on the list. Location is solid.
3BR on Reines Street — ₪9,800/mo, renovated, North TLV. The “special listing” energy.
Old North 3BR Mamad (new building) — ₪9,350/mo. Good area, newer construction.
Epstein Street 3BR — ₪11,000/mo. Established address, consistently in demand.
Kochav Hatzafon 4BR/150sqm + 55sqm terrace — ₪13,000/mo. This is the dream setup, huge outdoor space.

Three of these are within budget. Two push the top of it but come with outdoor space or extra rooms that change the value calculation entirely.

The Part That Would Have Taken Me an Hour

Each of those five listings — the contact, the location verification, the price-per-sqm math, the “is this area actually good?” judgment call — that takes a human roughly 15-20 minutes per listing. Five listings. You’re at an hour and a half before you’ve even scheduled a viewing.

With the AI layer running: under three minutes, zero manual filtering, and the rankings are based on your specific criteria — not what some real estate agent wants you to see.

What This Actually Changes

The goal was never to “find apartments faster.” It was to remove the apartment hunt as a persistent mental load — the background process that runs all day and steals attention from everything else.

When you can say “find me something” and get back five qualified options with price, location, and links — in under five minutes — apartment hunting stops being a hobby and becomes a transaction. Much better.