The 15-person AI-native services company. Its name is Every.
by Ygor Fonseca, Founder & Systems Lead
Every employs 15 people. They publish a daily AI newsletter, ship five products, run a $1–2 million-a-year consulting business, and write 100% of their code with AI. Their subscription business sits at $1.2 million ARR growing 15% month-over-month. The consulting arm is booked through Q1 2026.
That's the shape of a small AI-native services-and-products company. It exists today. It is small enough to copy and detailed enough to study.
What 15 people now ship
Every (every.to) is a media company that figured out how to be a software company at the same time. The structure is the part worth looking at closely.
Five products, one general manager each. Cora, Spiral, Sparkle, Lex, and Monologue — each run end-to-end by a single GM. The GM holds the roadmap, the customer relationship, the pricing, and increasingly directs AI agents to handle the engineering work. The Lenny's Newsletter podcast walks through the operating model: the GM-per-product structure only works because AI tooling does most of what a small product team used to do.
A daily editorial output that doesn't require an editorial team. The publication is high enough quality that subscribers pay for it; the team running it is small enough that in a 1995 magazine economy, the math wouldn't have been possible. AI tooling closed the gap between what a tiny editorial team can produce and what a paying audience expects.
A consulting arm that is also small. Every's Master Plan calls it out: ~$1–2 million in annual revenue, working with companies on how to actually integrate AI into their operations. The same staff who run the products and the editorial run the consulting. What they sell is what they have already learned doing the rest of the work.
An internal memory layer that ties it all together. Part II of Every's Master Plan — the most underrated AI-native essay of the last year — describes a memory layer that lets context travel across all five products, the publishing system, the consulting work, and the team's internal communications. Customer context for one product flows into the next. Editorial themes inform product decisions. Consulting engagements feed back into the publishing. The memory layer is what makes 15 people behave like a team three times that size.
Why this shape works
The Every architecture is not a special arrangement. It is the early-mover version of what most small services companies will look like by the end of 2027, and it works because of three structural choices.
One accountable human at every customer-facing surface. The GM for each product is the named operator. The consultant on every engagement is the named operator. The newsletter has named bylines. AI handles the production work — but the customer is always in a relationship with a specific human. This is the same architecture that Crosby, Harper, Basis, and the other AI-native services companies all share. The technology lets the team be small. The named operator is what lets the customer trust the work.
Working on AI to demonstrate AI. Every publishes a podcast called How Two Engineers Ship Like a Team of 15 With AI Agents. The transcript is publicly available. The point is not the headline number; the point is that Every uses every public surface — the newsletter, the podcast, the products themselves — to show how the operator stack actually runs. The transparency is the marketing.
Subscription revenue that funds the bets. 15% month-over-month subscription growth on the editorial side gives Every the runway to ship five products, run a consulting business, and build the memory layer underneath all of it. They are not raising venture rounds to subsidize the experiment. The publishing is paying for the building.
What this means for everyone smaller than Every
Every is 15 people. The shape is studyable. The architecture is documented. The question for any services owner reading this is the one Block's "From Hierarchy to Intelligence" essay puts at the center — what does your company understand that is genuinely hard to understand, and is that understanding getting deeper every day?
For Every, the answer is structural: a memory layer that makes the team's understanding of their customers and their products compound across every surface they operate. For most services companies today, the answer is "nothing" — the team's understanding lives in five Slack channels, three Google Docs, and the heads of four senior employees. AI cannot help a company that has not made its understanding queryable. AI can transform a company that has.
The most important takeaway is not the headline number — 15 people doing the work of 50. That is true, but it is not the lesson. The lesson is that the headline number is only possible because the operating model was rebuilt around AI from the inside, not bolted on top of the existing model. Every's structure — GM-per-product, daily editorial cadence, consulting arm, memory layer — is not a traditional magazine with AI tools added. It is a different shape entirely.
That shape is what AI-native services companies look like — in marketing or any other category — when the architecture is built right. Leanboat applies the same shape to growth marketing for consumer and software brands: AI does the production and the synthesis, named operators hold the customer relationship and the judgment, the work compounds because the system is built to let it.
If you want to map this onto your business, the place to start is the question Block asks. Then the audit answers it.