The web was built for humans, not for AI.

Every team that reads the web builds the same thing before it builds its own product, and then maintains it forever. Lyrenth is that thing, already built and already running.

No crawler to run. No fleet of headless browsers, no proxy pool, no retry ladder, no queue to babysit.

No parser for every layout. A page arrives as content, structure and provenance in one shape, whatever the site did to its HTML this week.

Nothing of yours breaks when a site changes. When a layout moves, it moves against our extractor on our machines. Your code is handed the same document shape it was handed yesterday.

One fetch of a page serves every reader. The first read brings a page into the index. Every read after that, yours or anyone else’s, comes from the index in milliseconds, and the site is not contacted again. A read is a read; what the index changes is the speed, and that a page a thousand people want is asked for once.

Nobody to apologise to. Our crawler identifies itself, follows robots.txt, publishes its IP ranges and answers site owners who write in. Your product is not the name in someone else’s abuse log.

Behind that sits an index of 4.1 billion documents across 171.7 million domains, and it is larger by the time you finish reading this page. None of it is a feature list. It is work you no longer own.

Lyrenth exists because the existing web needs to become AI-readable. Not replaced. Not closed. Not owned by one model company. Transformed into a cleaner, structured, machine-readable layer that AI systems can use reliably.

Most public webpages are designed for human eyes. They carry menus, ads, scripts, popups, tracking code, styling, repeated elements, and noise. Humans ignore most of that automatically. Machines do not, so AI reads the web the hard way.

Lyrenth changes that by turning public URLs into clean, structured AIDocuments: machine-readable versions of webpages that are easier, faster, and cheaper for AI systems to consume.

One URL. One clean document. One canonical version every agent can read without rebuilding the same pipeline.

One URL is a document. The whole public web, cleaned this way, is an index: the layer AI reads and searches through instead of scraping.

Adopting infrastructure is not buying a feature. It is taking on a dependency, and the real question is whether the thing under you will still be there in two years. Here is what that rests on.

Lyrenth is bootstrapped, and it is built and run by its founder. That is not a badge, it is the reason there is no clock above it. Nobody is waiting on a return that forces a price rise. No board is pushing a pivot into something easier to sell. There is nobody above me who can decide to close the API you built on.

The crawl fleet, the index, the API and this page are run by one person. When a site owner writes to complain about our crawler, it reaches me and not a ticket queue, and I can stop the crawl on a domain the same day. I have done it, more than once.

It began at the end of 2025, with a short post I read while scrolling.

The next big thing is not only making AI smarter. It is giving AI the ability to see the web the way humans see it.

Until then I was building software, projects and businesses, and not thinking about internet infrastructure. After that I could not unsee the problem: every agent, assistant, crawler and model company was reading the same human web in its own way, paying for the same work again and again. The web already had the information. AI did not have the layer to read it. So I started building one.

The ambition is large because the problem is large. If AI is going to use the open web every day, then the open web needs an AI-readable layer: not thousands of private scraping pipelines, not every agent attacking the same origin server, not every company wasting tokens on the same messy pages.

Make the public web readable by machines.

That means building infrastructure that lets AI systems access web content in a cleaner, more reliable, and more efficient format.

For AI developers, it means less token waste, less latency, and less parsing.

For website owners, it means fewer redundant scrapers hitting the same pages.

For the internet, it means a better bridge between the human web and the AI systems now depending on it.

Lyrenth is not trying to replace the web. Lyrenth is building the layer that helps AI understand it.

There are two parts to this work.

AIWebIndex

The open standard for AI-readable web documents, stewarded at aiwebindex.org. It exists so the format does not depend on one company. Open for others to implement, inspect, and build on top of.

Lyrenth

The commercial implementation: the production index, API, crawler, infrastructure, tooling, and delivery layer that makes the AI-readable web usable today.

AIWebIndex defines the direction. Lyrenth builds and operates the infrastructure. We keep them separate because the standard should be bigger than one company, while the company competes on execution, scale, reliability, and trust.

Lyrenth is operated by Aleksma Ai, Inc., a Delaware corporation, building global infrastructure for the AI-readable web, starting with a large-scale index of public URLs transformed into clean, structured AIDocuments.

A clean document for one URL is useful. 4.1 billion of them, structured, deduplicated and ranked, become something else. They become an index an AI can search, not only read.

Humans have a search engine built for the way they browse. AI needs one built for the way it reads.

So the index is not only a place to clean a page you already have. It is where an agent finds the right pages in the first place.

The goal is not a small tool or a new feature. The goal is a new layer of the internet, and it is being built in the open against one of the biggest infrastructure questions of the AI era.

If AI is going to use the web, what should the web look like for AI?

You do not have to wait for the answer to use what is already built. One call, one clean document, today.

Aleksandar Martinovic

Founder, Aleksma Ai, Inc.

Aleksandar is an entrepreneur, private and public market investor, and cross-industry builder focused on building the search index for the AI-readable web.