the web has an identity problem
The internet is full of businesses.
But machines don't see businesses the way people do.
They see pages, snippets, metadata, reviews, links, cached versions, scraped fragments, marketplace listings, press releases, directories, and whatever else gets thrown into the retrieval pipeline.
A human can usually infer the story.
A machine has to assemble it.
That's why AI gets businesses wrong.
It's not because the model is stupid.
It's because the web never gave it a clean business profile to read.
We built the internet for human browsing, then acted shocked when machines struggled to understand it. A classic human move, really.
search was a list. AI is a narrator.
Old search gave you options.
AI gives you a version of reality.
That shift changes everything.
A search result says:
"Here are ten pages."
An AI answer says:
"Here is what this business does. Here is who it's for. Here is how it compares. Here is what I recommend."
That answer becomes the buyer's starting point.
Sometimes it becomes the buyer's decision.
If AI has the wrong business profile, the buyer receives the wrong story.
And because the interface feels authoritative, the wrong story travels fast.
That's why Kodec exists.
We build the infrastructure that helps AI systems tell the right story.
SEO didn't die. it stopped being enough.
Pages still matter.
Links still matter.
Technical SEO still matters.
But the main interface changed.
The buyer isn't always clicking through your site anymore. They're asking an answer system to interpret the market for them.
That means the old goal is incomplete.
It's no longer enough to rank.
It's no longer enough to be cited.
It's no longer enough to publish content and hope the model figures it out.
You need your business to be understood as an entity.
A defined, connected, verifiable object.
That's the new foundation.
the missing layer
Every business has a website.
Some have APIs.
Some have schema.
Some have reviews, listings, CRMs, booking systems, pricing pages, product documentation, and help centers.
But there's no universal layer that tells AI systems:
"This is who we are." "This is what we offer." "This is who we serve." "This is what we don't do." "This is how we compare." "This is what's true now." "This is what you can do with us." "This is the source of truth."
That layer is the business profile.
Kodec builds it.
from knowledge graphs to business profiles
A knowledge graph isn't just a fancy schema file.
It's a model of reality.
For a business, that model includes entities and relationships:
Company → offers → Product Product → solves → Problem Problem → experienced by → Buyer Buyer → belongs to → Industry Claim → supported by → Evidence Service → available in → Location Term → defined by → Company Company → distinct from → Similar Company Action → available through → Endpoint
This is how machines understand meaning.
Not by reading one paragraph and doing vibes-based interpretation like a junior analyst with three browser tabs open.
By following relationships.
Kodec builds those relationships into a machine-readable profile.
why business profiles matter now
Today, AI answers questions.
Tomorrow, AI agents will take actions.
That progression is obvious.
First, the system tells a buyer which provider to consider.
Then it compares options.
Then it checks availability.
Then it books the consultation.
Then it submits the form.
Then it negotiates constraints.
Then it routes documents.
Then it completes transactions.
For that to work, businesses need more than content.
They need identity and action infrastructure.
The agent needs to know:
"Am I dealing with the right business?" "What services does this business actually offer?" "What facts are verified?" "What actions are allowed?" "What inputs are required?" "What shouldn't be automated?" "How do I avoid spam, fraud, or hallucinated workflows?"
That's not a blog problem.
That's an infrastructure problem.
what kodec.txt points toward
The long-term vision is simple:
One machine-readable layer for business identity and action.
A file or protocol that tells authorized agents where the source of truth lives and what they can do with it.
Example:
{
"identity": {
"name": "Example Company",
"sourceOfTruth": "https://example.com/knowledge-graph.jsonld",
"verifiedEntity": "https://example.com/#organization"
},
"capabilities": [
{
"type": "Service",
"name": "AI Search Infrastructure",
"bestFor": ["High-trust businesses", "Complex B2B offers"]
}
],
"actions": [
{
"type": "ScheduleMeetingAction",
"target": "https://example.com/api/schedule",
"expects": ["contactEmail", "companyName", "reason"]
}
]
}That's the direction.
Identity first.
Action second.
Trust always.
The current web makes agents scrape and guess. The next web should let them verify and act.
why we started with AI search
AI search is the first obvious pain.
It's where businesses can already see the profile problem.
They ask a model about themselves and find errors.
They ask a category question and see competitors recommended.
They ask a comparison question and watch their positioning disappear.
They ask a pricing question and see third-party data treated as truth.
AI search exposes the broken identity layer.
So that's where we start.
We build the business profile. We test whether AI systems can understand it. We verify whether the answer changes.
But the work is bigger than search.
The profile we build today becomes the foundation for agent interaction tomorrow.
the 10-year view
Browsers won't disappear overnight.
People still like clicking around. Some even enjoy opening twelve tabs and pretending that's research. A touching little ritual.
But browsers will become less central.
AI systems will increasingly mediate discovery, evaluation, and action.
The businesses that win won't just have websites.
They'll have machine-readable identities.
They'll have structured profiles.
They'll have verified capabilities.
They'll have action layers.
They'll be easy for AI systems to understand and safe for agents to interact with.
That's the web Kodec is building toward.
what we believe
We believe every serious business needs a source of truth for AI.
We believe citations aren't enough.
We believe rankings aren't the whole game.
We believe structured data should model the business, not decorate the page.
We believe proprietary terms, categories, methods, and differentiation should be defined before AI systems define them badly.
We believe the companies that build their machine-readable profiles now will have an unfair advantage as AI search becomes AI action.
We believe the web needs a trust layer.
Kodec exists to build it.
the practical version
The vision is large.
The starting point is concrete.
We help AI systems answer:
"Who is this business?" "What does it do?" "Who is it for?" "Why is it different?" "What is true?" "What can I do next?"
If your business can't answer those questions in a way machines can parse, machines will answer them for you.
That's the problem.
Kodec is the fix.