How Stapilo is built, and why it's hard to rebuild

A non-technical look inside the product: what it contains, how the AI is woven through it, how it was designed for people who aren't technical, and the hundreds of decisions that make it work. Every figure comes from Stapilo's own code, measured on 25 October 2026.

1
One platform instead of a stack of tools

A coach or expert today typically pays for and connects eight to twelve separate tools: a page builder, an email tool, a webinar tool, a booking calendar, a course platform, a community, a checkout, a chat bot, an ads tool, video hosting. Stapilo is all of them, built as one system with one list of customers, so nothing has to be glued together.

Behind these sit 336 connected database tables. Every action a customer takes (a purchase, a webinar visit, a form answer, a message) lands in one history, which is what lets the AI understand a business as a whole.

2
The AI is the operator, not an add-on

In most platforms, AI is a helper bolted onto screens that were designed for humans. In Stapilo, the whole product is organised so that one AI agent can find its way around it and work in every part of it.

• The AI's map is the app's map. The rooms, tabs and sub-tabs a person clicks through are exactly the places the AI knows. Automated checks keep the two identical, so the AI never describes a screen that doesn't exist.
• One conversation per business. The whole team works with the AI in one shared conversation, in every room, so nothing said in one place is forgotten in another.
• It acts, it doesn't just advise. It builds pages, sets up webinars, writes and schedules campaigns, answers customers, and runs ads, through the same actions a person would use.
• It asks first. Anything risky (going live, sending to many people, deleting, refunding) waits for the owner's yes.
• It opens to outside AI. The owner's own AI assistant can operate Stapilo through a secure connection, with exactly the owner's permissions.

3
Designed for people who aren't technical

Stapilo's main customers are experts who are new to selling online. Every design choice aims at one thing: a person should be able to start without learning software first.

• Talk on the left, see it on the right. The person says what they want; the screen beside the conversation shows the result as it's built.
• Every area has the same shape. The same tabs (overview, people, analytics, reminders…) sit in the same places everywhere, so learning one area teaches all of them.
• Sold by the job. Single products and packages built around what a person wants to do, in three sizes as a business grows; a single tool can be bought on its own.
• Customers get a finished experience too. The coach's own customers sign in to one place for their courses, events, bookings and purchases, with a mobile app for communities.
• Every screen is translatable, and the whole product is built to be used in more than one language from day one.

4
Hundreds of decisions you can't see

What makes Stapilo hard to rebuild isn't the amount of code. It's the decisions inside it, each learned from a real situation. A few, in plain words:

Each of these is enforced by an automated guard: if any change breaks one, it can't be shipped. Stapilo has 573 such guard files and 3,173 automated test files (about 466,000 lines of tests) that run on every change.

5
Built to be safe and reliable

Stapilo runs as five separate programs across four servers. A stranger chatting on a sales page runs in a different program from a paying business's AI, so a problem in one can't take down the other. Pages and mini-apps that businesses build run on separate addresses, sealed off from accounts and data. Security audits are run on the live system, not just the code.

6
Connected to what businesses already use

Still missing, stated openly: Zapier/Make connections and a public API with keys are on the build list.

7
The effort, quality and security, measured

Measured from the code history for February–October 2026 only (the earlier months of development were in a previous codebase). Every figure can be reproduced from the repository.

Built and rebuilt, month after month

3.07 million lines written and 1.49 million removed in eight months: the product wasn't generated once and left alone. Whole parts were rebuilt when a better design was found. In September alone the old dashboard and the old AI engine were replaced.

A real team, working through a disciplined process

Since June every change goes through a pull request and must pass the automated checks before it can be merged. Stapilo was built together with an AI coding assistant, practically all of it (94% of changes are recorded that way). That's what let a small team build this much, and every change passes the same tests and guard rules either way. "Person-days" counts the days each person recorded work in this codebase; the design, thinking and testing behind each day, and the earlier months of development in the previous codebase, come on top.

Security and data safety

Stated openly: a full restore drill from the off-site copies hasn't been recorded yet; it's on the list.

8
What it would take to rebuild

Stapilo is 18 months of development, at least 8 of them with a team of five, using AI throughout, which is what let a small team build this much. A competitor could copy individual features. Rebuilding the whole means rediscovering the hundreds of decisions above, building the AI around every area from the start, and keeping 573 guard rules and 3,173 test files passing while doing it. That takes a team and time, and it's exactly what established platforms find hardest, because they'd have to rebuild a system their customers already depend on.

← Back to the valuation · Price research · Customer model · Full feature comparison · Full technical reference