SWFL.com — The Connective Tissue
for an Entire Region.

The Situation

Southwest Florida is roughly 2.6 million permanent residents and another 150,000–250,000 seasonal snowbirds spread across six counties — Lee, Collier, Charlotte, Sarasota, Hendry, and DeSoto. Community life, events, and small-business commerce are scattered across Facebook Groups, Meetup, LinkedIn pages, a dozen chamber-of-commerce sites, university calendars, and old-school print magazines.

Nobody owned the unified discovery layer. National platforms (Meetup, Facebook, Yelp) are not regional. Local incumbents (chambers, magazines) are not digital-native. And the rise of AI search means the answer to "what's happening in Naples this weekend" will increasingly be given directly by ChatGPT, Claude, and Perplexity — pulling from whatever structured source surfaces best. The first platform to become that structured source for the region wins both the human traffic and the AI-mediated traffic at the same time.

What NURO Built

SWFL.com is a Next.js 15 + Supabase + Postgres/PostGIS platform that NURO designed, built, and shipped end-to-end. Live in production with daily ingestion across six counties:

The discovery layer

200+ organizations, 50+ businesses, 500+ events, and 81 curated knowledge-base entries — across dining, neighborhoods, schools, healthcare, parks, real estate, transportation, weather, shopping, cost of living, government, and beaches. Filterable by city, county, category, date, radius. An interactive Leaflet map with county and category toggles. Full-text search powered by Postgres pg_trgm, semantic search by pgvector with OpenAI embeddings. Per-county and per-city landing pages with FAQ JSON-LD for discoverability.

The AI Concierge

The product's sharpest edge. Embedded on the homepage and exposed via a standalone, rate-limited chat endpoint. Streams answers from Claude via OpenRouter, grounded in live events, organizations, businesses, and the curated knowledge base via tool-use. Intent classification routes city/county/category signals to the right surfaces; recommendations are bounded to what's actually in the database — no hallucinated venues. Premium subscribers get the Concierge plus a personalized weekly email digest driven by their preferences, neighborhood, and saved categories.

Daily ingestion across the region

The platform ingests events daily from the Eventbrite API, the Meetup public surface, and 25+ RSS and iCal feeds — chamber calendars, government calendars, universities, regional aggregators. Admin tooling extracts event details from URLs with AI fallback when feed metadata is thin. Submitted events and groups flow through a moderation queue; business listings can be claimed via auto-verify (domain DNS) or manual review.

A working monetization stack

Three monetized surfaces, all live: consumer subscriptions (Premium and a snowbird tier with calendar OAuth and the personalized digest); per-county business inclusion via a Managed Channel Placement model (businesses set a per-county bid and monthly cap to be surfaced by the AI Concierge, billed against tracked impressions and clicks); and free newsletter capture with downstream automation. Stripe handles billing end-to-end — checkout, webhooks for activation / cancellation / payment failure, and tier-specific feature gating.

CRM and lifecycle automation

Every contact event — signup, plan upgrade, business claim, org-admin approval, newsletter join, event submission — pushes a tagged contact into GoHighLevel via a canonical tag taxonomy (source, tier, category, county, role). That taxonomy is the integration contract: every downstream GHL workflow, campaign, and pipeline triggers off tags rather than IDs, which means the marketing team can change automation in GHL without touching the app. Transactional email — welcome, cancel, payment-failed, the weekly digest — runs through Resend with React Email templates.

Multi-tenant authentication and an admin moderation surface

Supabase Auth with email plus Google and LinkedIn OAuth. Row-level security on every table: public read for orgs and events, authenticated for saves and follows and Concierge, org-admin scope for org management, admin scope for approvals and taxonomy. Authenticated users save events, follow organizations, claim businesses, set agent preferences and calendar connections, and can opt into an @swfl.com email alias. The admin surface handles organizations, submission queues, claim queues, events, users, commercial accounts, MCP categories, and feed sources — with audit logging on the moderation actions.

AI discoverability built in from day one

The same principle NURO applied to its own site: be the structured source AI assistants actually cite. SWFL.com ships with llms.txt, /.well-known/ai.txt with bot-specific allow rules for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others, plus /ai/summary.json, /ai/faq.json, and /ai/service.json. JSON-LD schema (Organization, Event, LocalBusiness, FAQPage) on every relevant page. A dynamic sitemap and an RSS feed of the next 50 upcoming events. The platform was designed to be answer-engine-friendly from the first commit.

The Outcome

SWFL.com is live across all six target counties with daily ingestion running, all three monetized surfaces operating, CRM tag taxonomy enforced across every conversion event, and the AI Concierge answering grounded questions against live data. The infrastructure that took most regional platforms years to assemble — discovery, search, an AI agent that actually retrieves, a working ad model, a working consumer subscription, calendar OAuth, an admin moderation queue, a business claim flow, and a CRM contract — shipped as one coherent system.

Why It Mattered

SWFL.com is the proof that vertical-calibrated workflows are not a slogan. The platform is industry-specific and region-specific by design — the things that make it work (county taxonomy, snowbird-season modeling, chamber-feed ingestion, the per-county MCP ad model) are wrong defaults for any other product. That is the point. The platform was built around how this market actually behaves, not how a generic events product would assume it does.

The AI Concierge sits inside the same Truth Boundaries discipline NURO ships with every build: the agent is constrained to verifiable facts pulled from the live database — it cannot recommend a venue that does not exist or fabricate an event that is not in the calendar. The Decision Log keeps every Concierge interaction auditable. The result is a regional answer engine that an LLM can cite without embarrassing itself — and a platform that competes on structured-data quality before it ever competes on SEO.

And the broader pattern is one NURO runs across every engagement: the strategic consulting layer up front (what should the product actually be), the custom build sized to that answer (Next.js + Supabase + a real billing stack instead of a no-code lash-up), and the embedded delivery afterward (NURO continues to operate SWFL.com's ingestion, moderation tooling, and Concierge improvements, not a hand-off-and-disappear).