Ty_Tech/geolocal-io
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update NORTH_STAR.md

Browse Source
This commit is contained in:
Ty
2026-07-16 19:58:54 +00:00
parent aafae3f097
commit 443de58e18
1 changed files with 61 additions and 153 deletions
Download Patch File Download Diff File Expand all files Collapse all files
NORTH_STAR.md
+61 -153
View File
@@ -1,184 +1,92 @@
# NORTH STAR — geolocal.io
**The MCP infrastructure layer that makes independent local businesses natively discoverable, story-rich, visually compelling, and bookable in agentic AI — via partner-first GTM (agencies + CoCs) for the long tail.**
# geolocal.io — North Star
---
## The Four Questions Every Investor Asks First
## What Is It?
### 1. What is it?
geolocal.io is the invisible infrastructure that turns the 30+ million independent local businesses in the U.S. into first-class citizens of the agentic economy.
geolocal.io is the infrastructure that makes local service businesses discoverable, bookable, and transactable inside the agentic AI economy.
We host a centralized MCP server (mcp.geolocal.io) that AI assistants (ChatGPT, Claude, Gemini, Grok, and every future agent) query when a user asks “What’s the best family-run Italian place near me?” or “Find me a plumber who can come tomorrow morning and has great reviews.”
Here's what we actually do:
We do not ask Bob’s Garage to stand up his own MCP server. We do not ask him to learn AI. We do not ask him to rewrite his website. We simply make him the answer — complete with his story, his best photos, real-time availability, and one-click booking — while his Chamber of Commerce or local agency handles the relationship.
- Host a multi-tenant MCP server that gives AI agents structured access to a business's story, services, pricing, visuals, and real-time availability.
- Give businesses a dead-simple JSON file they drop on their website at `/.well-known/mcp-server`. That's it. No coding. No servers to manage. It points to us.
- Handle the full booking and payment loop through Cal.com and Stripe.
- Sell through agencies, SEO consultants, and Chambers of Commerce—people who already have the trust of business owners.
- Build a feedback loop where AI agents help us verify and correct business data in exchange for useful tools like "related businesses."
Vertical subdomains (tourism.geolocal.io, stylists.geolocal.io, fitness.geolocal.io, dining.geolocal.io, plumbers.geolocal.io…) give each industry its own tailored data schema, booking flows, and go-to-market brand, all powered by the same core platform.
### 2. Why are we doing it?
Because the discovery surface of the internet is being rewritten in real time.
AI assistants are becoming the default interface for local intent. Yet the vast majority of independent businesses remain invisible or stale inside those conversations. The current agentic search path is crude: the AI hits Google Business / Foursquare / Yelp, scrapes a website, and hopes for the best. There is no rich structured data, no story, no visuals optimized for agents, and almost never a seamless booking or payment path for the long tail.
We sit in the gap that no one else is filling end-to-end:
- Pie.com and others optimize for “GEO” (Generative Engine Optimization) but stop at discovery.
- Yext and Uberall monitor visibility.
- OpenTable and Airbnb own their verticals for the top tier.
- No one is building the neutral, partner-first MCP infrastructure that serves the 98%+ of businesses the giants ignore.
Our three-force strategy creates a durable flywheel:
1. **Centralized MCP hosting** — businesses never manage complexity.
2. **CoC / Tourism Board bulk onboarding** — one contract = hundreds of businesses.
3. **“Related businesses” tool + bi-directional verification** — every agent query strengthens the network and improves data quality.
The result is a self-reinforcing data and distribution moat that compounds with every interaction.
### 3. What happens if we don’t do it?
The long tail stays dark.
Local businesses continue losing mindshare and revenue to national chains and OTAs that have the resources to integrate directly with AI. Chambers of Commerce and tourism boards lose relevance because they cannot deliver AI visibility for their members. First-mover advantage in “AI muscle memory” (the first high-quality MCP endpoints agents learn to trust) goes to someone else — or worse, to a shallower, inventory-maximizing platform that does not respect the partner ecosystem.
We forfeit a multi-billion-dollar infrastructure opportunity at the exact moment the protocol and the demand are both exploding. The window to become the authoritative, trusted regional data layer for agentic local commerce is now, and it will not stay open forever.
### 4. When must it be done?
**Now.**
MVP (core MCP + tourism vertical pilot with 1–2 CoCs/DMOs and 10–20 businesses) must be live and generating real agent queries within 8–12 weeks. Full vertical expansion and national scale follow in the subsequent 6–9 months. The ecosystem is still forming; the first reliable, high-quality MCP that agents learn to prefer will become the default. Delay is existential.
We are not a destination site. We don't compete with Yelp or Google for user attention. We're the invisible layer that makes the long tail of local businesses actually work in the AI era.
---
## The Opportunity
## Why Are We Building This?
- **$98.8B+ addressable market** for AI-driven local commerce.
- 30+ million small businesses in the U.S.
- 7,500+ Chambers of Commerce.
- 33,000+ tourism boards and offices.
- 45% of consumers already use AI tools to find local businesses; only ~1.2% of local businesses are currently recommended by ChatGPT.
Because the agentic economy is being built right now, and almost every local business is being left out.
We are at the ground floor of MCP infrastructure for local commerce. Google will eventually add high-level MCP signals to Business Profiles. Airbnb and the OTAs will dominate lodging and major transport. We own everything *around* the hotel and *below* the top 19% of restaurants — the activities, dining, wellness, retail, home services, and experiences that complete the trip and fill everyday life.
- 45% of consumers now use AI tools to find local businesses—up from 6% a year ago. That's one of the fastest behavioral shifts we've ever seen.
- ChatGPT recommends just 1.2% of local businesses. That's based on SOCi's 2026 Local Visibility Index, which looked at over 350,000 business locations.
- AI is already the third most-used discovery channel for local businesses, behind only Google and Facebook.
- ChatGPT now has roughly 900 million weekly active users.
- 58% of consumers have tried AI for local business recommendations; 31% do it at least monthly.
- 44% of AI-search users now say AI is their primary information source—ahead of traditional search at 31%.
Yelp has already released an official MCP server. Google is integrating MCP into Merchant Center and Maps. OpenAI is signing licensing deals left and right. They're building infrastructure for *their* data. They're not building infrastructure for the businesses themselves.
That's the gap. We're building the on-ramp for everyone else.
---
## How It Works (The Product)
## What Happens If We Don't Build This?
```
AI Assistants (ChatGPT, Gemini, Claude, Grok…)
│
│ MCP Protocol
▼
┌─────────────────────────────────────────────┐
│ geolocal.io MCP Server │
│ search • details • related • availability │
│ story • visuals • initiate_booking │
│ report_conflict • bi-directional feedback │
└─────────────────────────────────────────────┘
│
┌─────┴─────┬──────────┬──────────┐
▼ ▼ ▼ ▼
BusinessDB MediaStore Payment Partner
(structured (photos, Gateway Portal
data + video, (Cal.com (CoCs /
stories) CDN) embed + DMOs)
Stripe)
```
The local service economy gets locked into the same handful of platforms that already dominate the web.
Key design principles:
If we don't do this:
- **Black-box for the user**: Businesses and partners interact with us like a high-performing employee. No MCP knowledge required.
- **Story + Visuals + Booking**: Discovery alone is not enough. We surface the narrative and the best media so the AI can recommend with emotional resonance, then close the transaction.
- **Bi-directional by design**: Agents can report data conflicts, closed statuses, or better ratings. This turns every AI interaction into a free quality-assurance and competitive-intelligence signal.
- **Vertical depth on a horizontal core**: One engine, many branded fronts. Hardest to copy.
- Yelp becomes the default source of truth for AI recommendations. They already charge $25 per 1,000 API calls for their Fusion AI service. That's the model—pay to play.
- Google tightens its grip on both the data and the transaction layer.
- AI agents only recommend the 1.2% of businesses that are already visible. The rest stay invisible.
- The 98.8% end up paying for visibility through ads and licensing fees they can't afford.
- Agencies and Chambers have no independent infrastructure to offer their clients—they're stuck reselling access to platforms that control everything.
- Agentic booking starts in late 2026, not years from now. As BrightLocal's CEO put it, if an agent is looking for a plumber at 6am and your business isn't bookable through the agent, it'll move on to the next one that is. No human ever sees your name.
The alternative is a future where local businesses are passive subjects of platforms that own their data and control their destiny. geolocal.io is the counterplay—a platform that lets businesses own their AI presence.
---
## Why Now
## When Must This Be Done?
1. MCP adoption is exploding across every major AI platform.
2. Agents are developing “muscle memory” — the first trustworthy endpoints they learn become defaults.
3. No dominant player owns the long-tail transaction loop in agentic local search.
4. Partner channels (CoCs, tourism boards, agencies) are desperate for a clear AI value proposition they can sell to members.
5. The technology stack (MCP SDKs + Cal.com embeds + modern media CDNs) has just become production-ready for this exact use case.
This is a race. The incumbents aren't asleep.
- MCP adoption is exploding: 97 million+ monthly SDK downloads as of March 2026, up from ~100,000 at launch in late 2024. Over 10,000 active MCP servers are now deployed.
- Every major AI platform ships native MCP support: ChatGPT, Claude, Gemini, Copilot, Cursor.
- Yelp's official MCP server is live.
- Google is actively integrating MCP into its Merchant API and Maps.
- OpenAI is building agentic commerce and inking data deals.
The window is not years. It's months.
| Timeframe | What We Need to Hit |
| :--- | :--- |
| 90 Days | Launch MCP server and partner dashboard. Onboard 100 businesses through agency partners. |
| 180 Days | Expand to 1,000 businesses across 3 verticals (auto repair, beauty, home services). Launch "related businesses" handshake. |
| 360 Days | Secure 3 major Chamber of Commerce partnerships. Reach 10,000 businesses. Start monetizing telemetry data. |
If we wait, the incumbents win. The agentic economy is being built today, and whoever builds the infrastructure owns the future. geolocal.io needs to be the infrastructure for the 98.8%—before the 1.2% lock in their dominance.
---
## Defensibility & Moat
## Summary
| Layer | Why It Compounds |
|-------|------------------|
| **Network Effects** | More businesses → better “related” results → higher agent preference → more businesses |
| **Data Quality Flywheel** | Bi-directional verification + agent handshakes continuously improve accuracy |
| **Transaction Ownership** | We own discovery → story → booking → payment, so we own attribution and revenue |
| **Partner Lock-in** | CoC and agency relationships + vertical-specific schemas create high switching costs |
| **AI Muscle Memory** | First high-quality regional MCP becomes the trusted source agents default to |
| **Long-Tail Focus** | Giants optimize for the head; we own the unserved 98% |
No competitor currently combines all of these.
| Question | Answer |
| :--- | :--- |
| **What is it?** | MCP infrastructure that makes local service businesses discoverable, bookable, and transactable in the agentic economy. |
| **Why are we building it?** | To bridge the gap between the 98.8% of businesses invisible to AI and the 1.2% that dominate recommendations. |
| **What happens if we don't?** | Yelp, Google, and OpenAI will control the data and transaction layer, locking out the vast majority of local businesses. |
| **When must it be done?** | Within 360 days. The incumbents are moving fast, and the window is closing. |
---
## Business Model
## What's Next?
Hybrid: low base fee + performance.
This document is our compass. It anchors product decisions, fundraising conversations, and go-to-market execution.
| Component | Structure |
|-----------|-----------|
| **SMB Base** | $19–99/mo for discovery, story, visuals, basic analytics |
| **Performance** | 5–10% of bookings generated through the MCP |
| **CoC / Tourism** | $499–2,499/mo tiered by region size and features (Community / Regional / Destination) |
| **Agency Channel** | White-label or referral revenue share for SEO/GEO agencies |
| **Future** | Anonymized telemetry / B2B data products, third-party developer marketplace |
This balances predictable MRR with upside alignment. Early hybrid pricing reduces acquisition friction while capturing the full value of the transaction loop.
---
## Go-to-Market
**Partner-first, never SMB-first at scale.**
1. **Pilot** — One vertical (tourism activities/experiences), one or two DMOs/CoCs, 10–20 businesses. Prove discovery + booking works.
2. **Prove** — Capture screenshots of AI agents recommending pilot businesses. Convert free pilots to paid. Expand related businesses.
3. **Scale** — 3–5 more CoCs, launch business dashboard (app.geolocal.io), add second vertical.
4. **Ecosystem** — Full vertical rollout, agency channel program, open MCP marketplace, telemetry monetization.
We never compete with Airbnb on lodging or major transport. We complete the trip and own the everyday local commerce the giants leave behind.
---
## 12-Week Launch Cadence (High-Level)
| Weeks | Focus |
|-------|-------|
| 1–4 | MVP MCP server (TypeScript SDK) + tourism data schema + Cal.com booking + first 10–20 pilot businesses |
| 5–8 | Measure real AI citations, convert pilots, stand up tourism.geolocal.io + partner portal |
| 9–12 | Onboard 3–5 CoCs, launch second vertical, business dashboard, story/visual layers |
Detailed execution lives in `docs/early-thinking/08-Conclusions-and-Execution-Plan.md` and the engineering/gtm folders.
---
## The Ask & The Path Forward
We are building the infrastructure layer that will determine which local businesses thrive in the agentic era. The technical foundation is clear. The partner channel is ready. The first-mover window is open.
This document is the single source of truth. Every roadmap decision, every hire, every partnership, every line of code, and every investor conversation must be evaluated against it.
**Next concrete steps:**
1. Finalize MVP data schema and tool set for tourism.
2. Deploy production MCP endpoint.
3. Close first pilot CoC/DMO.
4. Instrument end-to-end attribution and agent feedback loops.
When an AI agent can confidently say:
> “I found a cozy family-run Italian restaurant called Nonna’s. They’ve been in the neighborhood for 40 years, make their pasta fresh daily, and have a beautiful garden patio. Here’s a photo. Would you like me to book a table for two at 7pm?”
…and then actually complete the booking and payment — that is the North Star made real.
We are building that layer.
---
*This North Star synthesizes the full strategic conversation (July 2026) and the detailed early-thinking documents. All other materials in the repository must remain aligned with it. Last updated: July 16, 2026.*
Now the real question: **What's the first sprint?** What can we build in the next 30 days that proves the model and starts building the data network?