The Par AI unifies your property inventory into one live workspace, then puts an AI assistant on top that answers only from data your team published, it never invents a price or a unit. It reasons with the buyer, weighing budget, timeline and any unit specifics until they reach a decision that actually fits them. Swap your ad CTA for a Par AI link and every conversation arrives as a structured buyer brief.
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Maker
š
Hey Product Hunt š
I'm Amr, and I built The Par AI.
The problem that started it: a broker quotes a client a price from a screenshot in a WhatsApp group. The developer changed that price four days ago. Nobody noticed. Either the deal dies or worse, it doesn't, and someone has to walk it back in front of the client.
Most "AI for real estate" makes this worse, not better. If your assistant answers property questions from a vector store built off uploaded PDFs and brochures, it will confidently quote a price that was true last month. And in off-market and new-construction inventory there's no MLS to sanity-check it against the developer's own system is the only source of truth that exists.
So we built the inventory layer first, and put the assistant on top of it.
ā The assistant answers only from live, developer-synced inventory. There's no webhook to break and no embeddings to go stale ā the assistant and your inventory are the same system.
ā It doesn't just retrieve, it reasons with the buyer. It works through budget, timeline, household and must-haves with them, narrows the inventory down as the conversation goes, and helps them reach a decision that actually fits ā instead of dumping ten listings and hoping one sticks.
ā Import the spreadsheet you already use, with a preview before anything saves. You're not migrating off Excel, you're importing it.
ā Share units through tracked, expiring, password-protected links ā and see who actually opened them.
ā Swap your ad CTA for a Par link. The prospect has a real conversation instead of filling out a form, and your broker gets a structured Intent Brief ā budget, timeline, household, decision makers, objections ā before the first call instead of after the third.
What it isn't: not a voice receptionist, not a transaction or compliance tool, not a comps/valuation engine. The lane is narrow on purpose ā inventory truth, and qualification sitting on top of it.
Three things I'd genuinely like feedback on:
1. If you've built RAG over data that changes underneath you ā how are you handling re-embedding when the source updates? That seam is the whole problem and I haven't seen a clean answer yet.
2. If you're in real estate: where does your pricing data actually live right now? Honestly.
3. Does "the AI reads, it doesn't generate" land as a real distinction, or does it just read as marketing?
I'm here all day. Ask me anything, including the hard stuff.