How Clara knows where homes are

So a caller can say "the homes by Skyler Street" and Clara knows which ones they mean.

2026-09-29 · all companies, no switch · street details are ours, nicknames are Gera's

In one sentence: every listed home gets its street, nearby streets, neighborhood and 3 landmarks from Google Maps, saved with the home, and Clara reads them when she talks about it.

AppFolio listings page the homes for rent Listed home record copied every 15 min Google Maps streets + landmarks asks Saved on the home (refreshed every 30 days) Street: Ember Lane Nearby streets: Skyler St, Florida St, … Neighborhood + 3 nearest parks, schools, stores, hospitals Clara: calls, texts, email "364 Ember Lane is right by Skyler Street." Caller "the homes by Skyler?" Nicknames (Gera) "White Willow" = Ember names no map knows coming separately Google says no → alert same day
Only listed homes. Clara only offers what's listed, so only listed homes are looked up. Each home gets its own street, so Ember homes get Ember Lane, not the community's paperwork address. Saved, not live: Google is asked in the background, never during a call.

Where it stands

PieceStatus
Clara sees every listed home (Ember: all 4)Merged today
Google keyFixed today
Details filling in (Western Slope)2 of 18, rest over ~2 hours
On/off switch removed, all companiesBeing built
Alert when Google failsBeing built
Tests on real calls and streetsBeing built
NicknamesGera
More detail: Ember example, what went wrong, what it won't do

The Ember call

On 9/22 and 9/25 a caller asked for "the new Skyler subdivision." Clara said she couldn't find it and the caller asked for a person. Google puts Skyler Street about 200 meters from 364 Ember Lane, so once the Ember homes fill in, Clara connects the two.

Why nothing was saved until today

Google refused our old key on every request, and the code failed silently. A new key went in today, and a change is coming that alerts us the same day if Google refuses again.

What it won't do

Cost

About 2 Google lookups per listed home per month, a few cents in total.

The bigger picture: where Clara's facts come from

Street details are one of five sources. Each has one job, and one source wins on each kind of question.

The client's AppFolio their system of record 1. Listed homes price, beds, photos, specials, marketing text copied every 15 min 2. Who's who resident, prospect, vendor? properties, units, tours 3. Knowledge base company: fees, pets, hours property: lockbox tours, no pets here each listed home 4. Street details Google, saved, 30-day refresh owned by us 5. Nicknames "White Willow" = Ember owned by Gera, coming Clara's briefing, rebuilt every call, text, email she answers, books tours, makes guest cards → written back into the client's AppFolio
AppFolio and the intake doc are raw material. The knowledge base is cleaned up per client and is where Clara's rules live.
QuestionSource that wins
Fees, pets, rules, how tours workKnowledge base (company level, then property)
Which homes are available, rent, specialsListings page
Where a home is, what's nearbyStreet details (Google)
What locals call a placeNicknames (Gera)
Who's callingThe client's AppFolio

Known break: fees reach Clara from both the knowledge base and a direct AppFolio feed. The fix is to fill fee gaps in the knowledge base, then remove the AppFolio fee feed (waiting on the client's answers).

PropFlow Docs