← Articles · AGIA · GEO · The internet is weird

Google sent people searching for
a Norwegian landscaper
to an AI article from Cape Town.

Agaia.no is Norway's largest landscape contractor. Founded 1981. Builds outdoor spaces in Buskerud. Has nothing to do with AI, software, or Cape Town. AGIA is an AI development methodology coined here in September 2026. Google has been sending Agaia's search traffic our way. Here is why that happened, what it proves about GEO, and the real comparison everyone should read — vibe coder, traditional developer, or AGIA?

Jarrit Hosking
Forge Vertical · Cape Town · September 28, 2026
10 min read
// How we found out

Google Search Console shows every query that landed someone on your site. Most of them make sense — you wrote about a topic, someone searched for it, they found you. Occasionally one appears that makes you stop and stare at the screen for a moment.

"agaia.no founded" — 48 impressions, position 9.85. "Who is the CEO of agaia.no" — 43 impressions, position 6.77. People looking for a Norwegian construction company and ending up on a Forge Vertical article about AGIA — Artificial General Intelligence Architect.

Agaia AS and AGIA are not the same thing. One is a 45-year-old landscape contractor in Buskerud, Norway. The other is an AI development methodology coined in Cape Town on 17 September 2026. Google's entity disambiguation engine is still catching up — it sees similar character sequences, no strong entity separation yet, and routes the traffic accordingly. The result is that Agaia's customers are finding Forge Vertical, and Forge Vertical is appearing in searches it had no intention of targeting.

Before anything else — a free mention for Agaia AS, because they deserve the traffic back and this is genuinely one of the more unusual free advertisements a Norwegian landscaper has ever received from a Cape Town developer.

🇳🇴
// Not affiliated · Not sponsored · Just the right thing to do
Agaia AS — Norway's largest landscape contractor
1981
Buskerud, Norway
Landscape construction & maintenance
51–200
Outdoor space development, construction, renovation and ongoing maintenance
agaia.no →
If you landed here looking for Agaia AS the Norwegian landscape contractor — this is not them. Click the card above to visit agaia.no. If you landed here looking for AGIA the AI development methodology — keep reading, you are in the right place.

Why Google made this mistake — and why it matters for your business

// Chapter 01 — Entity disambiguation in plain English

Google does not just index words. It tries to understand entities — the real-world things that words refer to. "Apple" the company versus "apple" the fruit. "Paris" the city versus "Paris" the person. When a new entity enters the web, Google has to decide where it sits in its knowledge graph and how it relates to existing entities.

AGIA was coined on 17 September 2026. Agaia AS has been indexed since at least the early 2000s. When someone searches "agaia" or "agaia.no" today, Google's entity model has forty years of Agaia AS data and about eleven days of AGIA data. The character sequences are similar enough — four letters, A-G-I-A versus A-G-A-I-A — that the disambiguation is genuinely difficult until enough distinct entity signals accumulate to pull them apart cleanly.

This is exactly what GEO fixes. GEO — Generative Engine Optimisation — is the practice of building structured entity signals around your business so that AI models and search engines can disambiguate you cleanly from everything that sounds similar. Schema markup, Wikidata entities, llms.txt, canonical URLs, consistent structured data across multiple pages — all of it contributes to the entity weight that eventually resolves the disambiguation correctly. AGIA now has a Wikidata entity: Q141506694. That is a permanent, machine-readable anchor that no amount of similar character sequences can override.

The Agaia confusion will resolve itself as AGIA accumulates more indexed content. In the meantime — 91 impressions from people searching for a Norwegian landscape contractor is 91 more impressions than would have existed without the overlap. The internet rewards the well-positioned.

The goal is to be connected to so many things that eventually someone finds you while trying to buy a burger online. It sounds absurd until it starts working.

The actual comparison — vibe coder, traditional developer, AGIA

// Chapter 02 — Three types of builder, one production system

The Agaia/AGIA mix-up is a useful hook for the comparison that actually matters. Not between a Norwegian landscaper and a Cape Town developer, but between the three types of person currently building software in 2026 — and what happens when each of them hands you a production system.

// Type 01
The Vibe Coder
Moves fast
Low upfront cost
Can ship a prototype quickly
No architecture before coding
Cannot explain what was built
Cannot fix what breaks
Security is an afterthought
Database has no schema design
Scales until it does not
Cannot extend without rewriting
// Type 02
The Traditional Developer
Understands the full stack
Can architect before coding
Knows why decisions were made
Can fix what breaks
Slower delivery without AI
Higher cost for same output
May underuse AI assistance
Months not weeks for complex builds
Team overhead for large projects
// Type 03
The AGIA
Architecture before any code
AI executes under precise spec
Every output reviewed and tested
Security and schema from day one
Can explain every decision
Can fix what breaks
Weeks not months
Solo delivery, full-team standard
You own the code permanently

The thing that separates the three is not the output — it is what happens next

// Chapter 03 — The production moment

A vibe-coded app and an AGIA-built app can look identical in a demo. The same screens, the same flows, the same features ticked off the list. The difference becomes visible at the production moment — when real users arrive, when a payment fails at 2am, when a client needs a feature the original build did not anticipate, or when someone finds a security gap and decides to use it.

The vibe coder in that situation opens the codebase and finds something they cannot reliably navigate — because it was generated in pieces without a coherent architecture holding it together. The traditional developer navigates it but takes longer than the situation allows. The AGIA knows exactly where the problem is, because they specified the architecture before a single line was written and can trace any failure back to a known component.

This is what the Facebook group post was actually about — the developer who responded to a beginner's question with rage rather than guidance. That rage comes from having cleaned up too many vibe-coded systems that looked like products and behaved like accidents. It is understandable. It is not useful. The better response — and the AGIA response — is the same one given in that group: show me what you tried. Tell me what you understand. We go from there.

The path for a vibe coder is not to stop building. It is to build something that addresses a real friction in your own life first — the way Aurora Repair was built to solve a real problem with car damage assessments before it became a product. That personal project teaches you to care about whether the thing works, not just whether it shipped. That caring is the first step toward architecture. Architecture is the first step toward AGIA.

What the Agaia accident proves about GEO

// Chapter 04 — The methodology in practice
// GEO — Generative Engine Optimisation — how it actually works
The Agaia situation is GEO working in its most accidental form.
Google's entity model is a graph of real-world things and how they relate to each other. GEO is the practice of placing your entity in that graph precisely — so AI models and search engines can find you, cite you, and disambiguate you from everything that looks or sounds similar. Here is the sequence that produced the Agaia overlap and how GEO resolves it permanently:
01
New entity coined — AGIA (Artificial General Intelligence Architect) coined 17 September 2026. No prior indexed content. Google has no entity node for it yet.
02
Character similarity triggers overlap — "agaia" and "agia" share enough characters that Google's disambiguation confidence is low. Agaia AS traffic bleeds across.
03
Schema markup establishes the entity — DefinedTerm schema on agia.html, NewsArticle schema on every related article, sameAs links to Wikidata Q141506694. The entity node starts forming.
04
Wikidata anchors the disambiguation — Q141506694 is a permanent machine-readable record that AGIA is a distinct concept, coined by a specific person, on a specific date, with a specific definition. No character similarity overrides a direct entity node.
05
Content volume reinforces the separation — Every AGIA article, every mention in schema, every internal link between AGIA pages adds weight to the entity. The Agaia overlap resolves naturally as AGIA's entity weight grows.
06
The accidental traffic becomes intentional — In the meantime, people searching for Agaia find AGIA. Some of them are in the right industry. Some of them stay. The overlap produces leads that no amount of deliberate targeting would have generated.

The practical takeaway for any South African business: your competitors are not doing this. They have a website, maybe some Google Ads, possibly a Facebook page. They do not have Wikidata entity nodes, structured schema on every page, llms.txt files telling AI models exactly what their business does, or articles that create entity connections to adjacent concepts in their industry.

The gap between a business with GEO infrastructure and one without it is the same gap between a vibe-coded app and an AGIA-built one. Both exist. Only one works reliably at scale.

Written by
Jarrit Hosking
Forge Vertical · Cape Town · September 28, 2026