Short by design. The platform itself is on the common list; this is the leverage layer — the tools that need history, volume or an existing database to be worth anything — together with the in-design work aimed squarely at production.
Each capability is tagged with where the AI sits: AI-driven a model produces the output · AI-assisted a model does part of it, rules or the agent finish · AI-supported no model in the step itself, but it runs on AI-produced input, distributes AI output, or is the human gate over it · AI-developed no model at runtime — the capability itself was built with AI · Program a brokerage program or human service, not software.
Leverage on a book you already have
The capabilities that need an existing database, farm or listing count to pay off.
“My Comps” on every valuation AI-assisted
The agent overrides ALEX with six of their own ListingIds and the engine still runs the full adjustment math.
user_comp_valuation_data_service.py90-Day Market Activity AI-developed
An OSM map, a grouped activity list, and “changes since last night” for a farm area.
ALEX Deck Render AI-developed
Listing and pitch decks rendered to image sets and a packaged PDF, brand-neutral, with a stability gate on the geometry.
Cloud Run alex-deck-renderEight-copy sync with a 52-check gate AI-developed
One command propagates a change to every place the content lives — both hosting sites, both editor copies, PDFs, bundles, short URLs, VERSION stamp — then gates it.
sync_property.py --apply · verify_property.pyLofty CRM as the system of record AI-supported
Contacts, sends and activity fan out to the CRM timeline, so the pipeline lives in one place rather than in four tools’ separate memories.
lofty_crm_adapter.py · linkedin_crm_fanout.pyPrior-listing history on every property AI-assisted
Previous list dates, prior sale notes and expired or withdrawn history surface on the property itself — the “I sold this street before” conversation, with the record attached.
fetch_prior_listings() · prior-listing CDOM ruleSolepoint™, turned on your own practice Program
The same analysis the advisory arm runs for institutions, pointed at the agent: what position in this market is yours alone, and what the evidence says it is worth. It answers in days the question coaching normally circles for a year — and it needs a track record to work on, which is why it sits here rather than on the common list.
public/solepoint.htmlFarm-area intelligence that reruns AI-developed
Market Pulse, the activity map and demographics are all scoped to a saved geography, so the same farm can be run every month and compared rather than rebuilt.
market_history_report.py · subdivision_scope.pyOne-Drop Listing Package
Everything from the studio’s delivery. One input, one approval.
The pieces exist separately today: correct the images, size them, write the copy, build the site, build the flyer, queue the emails. The seam between them is the work, and it is the part that gets skipped late on a Tuesday.
The specified pipeline takes the folder the photographer sent, exactly as it arrived — mixed stills, HEIC, floor plans and video — and returns the processed image set cut to every size, AI copy grounded in those actual photos, the published website, the print-ready flyer, the PDFs and the four-campaign email series queued as drafts. Confirm the address, then approve. Nothing sends or publishes on its own.
For a producing agent this is the difference between a listing that goes live the night the photos land and one that waits for a free evening. The marketing stops competing with the selling, and the site, flyer and emails cannot disagree with each other because all four are built from one manifest.
- • One drop in, seven artifacts out
- • Confirm the address, approve the result
- • No renaming, no re-uploading
- • Status: specified, not yet built
ALEX Geofarm
Draw the territory. Get the households inside it, ranked by who is most likely to move.
Farming fails for a boring reason: the list is a chore. The specified tool makes drawing the shape the whole job — every parcel inside the polygon comes back from the county record with owner, tenure, stories, living area, assessed value, exemptions and estimated equity, joined to MLS history and the current ALEX valuation.
Sort and stack on years owned, value band, absentee owner, never-listed, previously expired. One canonical list per farm with saved views rather than three drifting lists — so suppression, opt-outs and de-duplication happen once. Phone and email come from a skip-trace vendor, and no export leaves without a DNC scrub.
For a producing agent it turns farming from a monthly chore into a monthly diff: re-run the same shape and see what changed — new owners, new listings, tenure crossing a threshold. That change list is the message.
- • Draw it, don’t build it
- • Ranked, not alphabetical
- • One list, three saved views
- • Status: specified, not yet built