Why We Built Pilot

Pilot didn't start as a product idea. It started as two books about what's going wrong in the content industry — and what would have to be true for something to go right.

Latent Vector: AI and the Unlocking of a $5 Trillion File Cabinet argues that the most valuable knowledge in any organization is locked inside documents that nobody uses. Not because the knowledge is bad, but because the systems that hold it — CMS platforms, file servers, SharePoint sites, PDF archives — treat knowledge like files. They store it, they retrieve it, and that's all they do. The "latent vectors" of the title are the hidden dimensions of institutional expertise that traditional systems can't access or activate.

About Five Years: How Content Businesses Will Navigate the Next Half-Decade of Disruption maps the forces that are breaking the content business model. Three legs held it up for twenty years: search engines sent traffic, advertising monetized it, and human production costs were justified by the revenue. All three legs are failing at the same time. Search engines answer queries instead of sending clicks. Advertising yield declines as attention fragments. And AI collapses production costs toward zero for everyone, which means volume is no longer a competitive advantage.

Both books arrive at the same conclusion: the organizations that survive won't be the ones that produce more content faster. They'll be the ones that have something worth saying — accumulated knowledge, earned expertise, institutional memory — and can get that knowledge working across every surface where their audience lives.

Pilot is what that looks like as software.

The three-legged stool

The content business model's three legs — search traffic, ad revenue, and human production — aren't weakening. They're structurally failing.

Search disintermediation means Google answers the question directly instead of linking to your article. Your ten years of published expertise trains someone else's model, and the click never comes. The traffic leg isn't wobbling — it's being removed.

Advertising yield erosion means the attention that remains is worth less every year. Privacy regulation kills tracking. Platforms capture the ad spend. Publishers get the remnant inventory. The revenue leg doesn't support what it used to.

AI-driven production cost collapse means anyone can generate an acceptable article for fractions of a cent. When your competitor can produce the same volume at near-zero marginal cost, your production team isn't an advantage — it's an expense.

Most content businesses are responding by producing more content faster and cheaper. This is exactly wrong. Pilot was designed for the opposite approach: make the knowledge you already have work harder, across more surfaces, without adding headcount.

The schema trap

Latent Vector identifies what it calls the schema trap — the assumption that structured data and rigid taxonomies are the right abstraction for AI-native systems.

Traditional CMS platforms organize content by schema: articles have titles, slugs, body fields, categories, tags, and publication dates. This works for storing and retrieving articles. It doesn't work for understanding what an organization knows.

An article about construction safety and an article about building code compliance might share 80% of their underlying knowledge, but a schema-based CMS doesn't know that. They're two rows in a database with different tags. The knowledge relationship between them is invisible.

Pilot's knowledge base doesn't use schema as its organizing principle. It reads your documents, identifies topics, and maps relationships between them — not by metadata, but by meaning. Two documents about related subjects are related in the knowledge base whether anyone tagged them that way or not. This is what makes it possible for Pilot to write new content that draws from multiple sources coherently, rather than just retrieving one matching document.

Pull-to-push

About Five Years describes the shift from pull to push as one of the defining transitions in content distribution.

Pull is the old model: someone searches for something, your article appears in results, they come to your site. The reader initiates. You wait. Pull worked when search engines sent traffic reliably.

Push is the new model: content reaches people through feeds, newsletters, social platforms, and algorithmic surfaces. The content goes to the audience instead of the audience coming to the content. Push requires format fluidity — the same knowledge rendered as a tweet, a newsletter paragraph, a LinkedIn post, a web article, depending on where the audience is.

A traditional CMS doesn't support this. It publishes to one format on one surface: a webpage. If you want a newsletter, you rewrite the article as a newsletter. If you want a social post, you summarize the article by hand. Every surface requires separate human effort.

Pilot was designed for pull-to-push from the start. The knowledge base is the single source. Channels are the surfaces — web, email, social, PDF, CMS webhook. One piece of knowledge, many outputs, each shaped by the destination's format and voice settings. You configure the channel once. Pilot handles the format adaptation.

Format fluidity

Format fluidity — the idea that knowledge should flow freely between formats rather than being locked into one — is a recurring theme in both books.

The locked format is the article. Most CMS platforms treat the article as the atomic unit of content. Everything starts as an article and stays as an article. If you need it in a different format, you do the conversion manually.

Pilot treats knowledge as the atomic unit, not the article. The article is one format. A newsletter issue is another. A social post is another. A research brief is another. They're all projections of the same underlying knowledge, shaped by the channel's format constraints and the voice configuration active for that channel.

This is why Pilot can publish the same knowledge as a 1,200-word web article and a 280-character social post without someone rewriting it. The knowledge didn't change. The format did.

The information-commercial complex

Latent Vector uses the term "information-commercial complex" to describe the tangle of systems, vendors, formats, and workflows that organizations build around their content operations. CMS, DAM, email platform, social scheduler, analytics suite, SEO tool — each one stores a partial view of what the organization knows, none of them talk to each other effectively, and the organization pays for all of them.

Pilot isn't trying to replace every tool in the stack. But it does collapse several functions that traditionally required separate systems: content creation, knowledge organization, editorial voice control, multi-channel publishing, and source citation. These aren't separate problems that need separate tools. They're aspects of one problem — getting institutional knowledge from where it lives to where it's needed — that Pilot addresses as a single system.

The $5 trillion file cabinet

The title of Latent Vector — "AI and the Unlocking of a $5 Trillion File Cabinet" — refers to the aggregate value of institutional knowledge sitting unused in organizations worldwide. Not because it's worthless, but because the systems that hold it can't do anything with it beyond storage and retrieval.

That's what Pilot is for. Not to generate content from the internet. Not to be a better chatbot. To take the specific, hard-won knowledge that your organization has accumulated — in documents, in research, in archives, in institutional memory — and make it work. As articles, as newsletters, as answers to questions, as social posts, as whatever format your audience needs, in whatever voice your editorial judgment demands, with every claim traceable to a source document you uploaded.

The file cabinet is open. Pilot is what comes next.