Social scheduler built for real estate listings

The idea

Every listing generates the same social campaign: just listed, open house this weekend, price improved, under contract, just sold. Agents rebuild it by hand for every property on every platform, or pay a VA to do it badly. Postiz’s $186k MRR proves scheduling is still a paying market despite dozens of incumbents — differentiation wins slices — and the real estate slice has structure no horizontal will model: posts derive from a listing record and follow a lifecycle.

The wedge is paste-a-link-get-a-campaign: scrape the listing once, generate the whole week, and regenerate on every status change. The fair-housing copy check is the moat-flavored detail — agents are personally liable for steering language, no generic scheduler will ever build that check, and brokerages will mandate a tool that has it.

Who pays

Agents and small teams at $29-59/mo per seat — below the felt cost of one VA hour a week, paid from a marketing budget that already exists. Brokerages are the multiplier: a 30-agent office buys seats to standardize branding and reduce fair-housing exposure. Distribution through realtor Facebook groups and Instagram agent-coaching audiences, plus brokerage marketing directors who can sign 30 seats at once.

MVP scope

• Listing import by URL with photo, price, and spec extraction
• Five lifecycle campaign templates generating a scheduled post sequence
• Claude-generated platform-specific copy with editable voice profile
• Fair-housing language checker blocking flagged captions until fixed
• Publishing to Instagram and Facebook with a review calendar
• Skip for v1: direct MLS/IDX feeds, TikTok and YouTube, video generation, DM management, listing websites

Build ListingLoop, a social scheduler for real estate agents.

Core flow: an agent connects Instagram and Facebook via the Meta Graph
API (LinkedIn optional), then pastes a listing URL from Zillow,
Realtor.com, or their brokerage site. A scraper pulls photos, price,
beds/baths, square footage, and address into a structured listing
record the agent confirms. The agent picks a campaign template — Just
Listed, Open House, Price Change, Under Contract, Just Sold — and the
Claude API generates platform-specific copy for each post in the
sequence, using the listing specs and the agent's saved voice profile.
A campaign is a pre-scheduled week: teaser, full listing carousel, open
house reminder, day-of post. The agent reviews the queue on a calendar,
edits inline, and approves. Every generated caption runs through a
fair-housing checker: a rules layer flagging protected-class and
steering language (family-friendly, safe neighborhood, walking distance
to church) with suggested rewrites before anything can be scheduled.

Stack: Next.js App Router with TypeScript, Meta Graph API for
publishing, Claude API for copy and the fair-housing pass, Postgres via
Prisma with a scheduled-jobs worker.

Pages: listings library, campaign builder, content calendar, connected
accounts, brand/voice settings, billing.

Non-obvious details: publish via scheduled server jobs with retry and
token-refresh handling since Meta tokens expire mid-campaign; when a
listing status changes, offer one-tap generation of the next lifecycle
campaign instead of making the agent start over.

Pricing: $29/mo solo agent, $59/mo per seat for teams with brokerage
branding locks.