Repositioning deco

Rafael Crespo

August 3, 2026

deco is entering a new chapter, and over the last few weeks we've gone all-in on a repositioning: we've stopped selling ourselves as a website platform and started positioning as an AI-Service Company that operates and grows commerce for large brands. This post is about how we're doing that in practice, what's been painful, and what we're learning. If you're positioning or repositioning your own product, I hope it helps. And if you're a deco customer or partner, this explains what you'll see change.

What repositioning actually means

Positioning is the context you create to be understood: which market you're playing in, against whom, and why you're the best option in the world at something your ideal customer actually values.

Repositioning is changing that. It sounds simple, and it isn't, because the market's mind has inertia. People keep the first label they put on you, and they'll hand it back to you for years, even after your business has become something else.

Why deco needed to reposition

deco was born in 2022 as a platform: an open-source framework, a visual CMS, an edge runtime. We positioned as "the platform for building high-performance digital experiences" and grew that way, running some of the largest storefronts in Brazil.

The problem: to this day, a lot of people still see us as "a CMS" or "a frontend alternative." That label shrinks us twice. First, because it competes in a software-license market, which is a fraction of the money brands spend around their site (agency, development, upkeep, SEO, CRO). Second, because it describes the tool, and what our best customers actually bought from us was never the tool. It was the outcome: the faster site, conversion going up, the landing page live the same day.

The AI shift made this obvious. When agents can actually operate a store, finding where it's losing revenue, proposing the fix, shipping the change once it's approved, measuring, and learning, selling the tool for someone else to do that work becomes the wrong business. The right business is delivering the finished work.

And there's a lesson we share with PostHog, who inspired this post: we'd already tried repositioning gradually. We updated the site! We wrote posts! We changed the pitch! But never all at once, never with the whole team pointing the same direction. Incremental repositioning is like painting a house one brick at a time: nobody ever sees the new house.

How deco positions itself now

deco is an AI-Service Company for enterprise commerce. We operate and evolve the digital experience of large retailers, using AI agents, our own platform, and engineers, and we get paid on results, a percentage of GMV.

We're going all-in on this in an uncomfortable way. It's on the site, in the pitch, in the contract, in the content, in the videos, at events. And it will confuse some people for a while. It might cost us leads who were looking for "a CMS." We accept that short-term pain because our long-term success depends on one conviction: brands don't want to adopt AI, they want the finished result.

The vocabulary you'll see us repeat until you're sick of it:

• AI-Service Company is what deco is. A company that looks like a service from the outside and is software on the inside. You buy the outcome; we use our agents and our platform to deliver it with software margins and software scale.
• Software-Operated Commerce is the category. Commerce operations moving from a project with a start and an end into a continuous operation, run by software under human command.
• Agentic loop is the unit of work. Observe, diagnose, propose, a human approves, execute, measure, learn. That's how work happens at deco, and it's the opposite of a chatbot waiting for a prompt.
• deco Studio is the platform all of this runs on: the agents, the CMS, the task board, the orchestration. Open source, MIT licensed, self-hostable if you want, with your own LLM keys if you want.
• Diagnostic is the entry point. You paste your store's URL and get 389 checks on where it's losing revenue. Free, nothing to install.

A huge part of positioning comes down to picking a handful of specific words and repeating them everywhere until your own team is sick of them. By the time your team is sick of them, the market is just starting to notice.

If it's not the CMS market, what market is it?

Our real competitor isn't another platform. It's the old way of working: the ticket waiting for a sprint, the agency's endless backlog, the replatform every two years, the dashboard that points at the problem without fixing it, the hourly contract that pays for effort instead of results.

Against that, we compete for the services budget, which is several times larger than the software budget. And we're not alone in seeing this shift: Sequoia calls it "services as software," a16z talks about the palantirization of software, Emergence Capital shows that for every dollar spent on software, companies spend several more on services. The thesis is out there. What we're doing is applying it, for real, to commerce operations.

Why coin new terms?

Because we're in a market that doesn't have an established name yet, and that's category-creation territory. It's different from ordinary positioning, where you differentiate inside a market that already exists. Category creation is fun, risky, and hard: you're trying to change people's mental model of how something established should work. Sometimes that requires a new word, because the thing doesn't exist in the vocabulary yet.

The rule we follow: when a term is already an industry standard, we use that term. We're not going to redefine "MCP" or "agent." But "AI-Service Company" applied to commerce, and "Software-Operated Commerce," describe something that still has no consensus name, so we take on the work of teaching it. It'll look like we're repeating ourselves too much. That's on purpose.

Are you open to changing your mind?

Someday, maybe. Right now, no. Internally we've already gone through enough rounds of debate, and everyone had to disagree and commit for this to leave the page. Arguing about a category name is the perfect way to distract an entire company for months.

Part of repositioning is committing for long enough to know if it's sticking. The early signs are good: new contracts are already being signed on the results model, customers who knew us as a platform are expanding into operation, and the conversation has shifted from "how much does the license cost" to "how much will you grow my store." But most of the market still calls us a CMS. It'll take time. We'll get there.

As always, we'll know if we need to correct course the same way we decide everything here: by talking to customers and building what they want.

And a reading recommendation

There are a thousand books on positioning and every one of them has some flaw. The one that influenced us most was "The 22 Laws of Category Design," from the Category Pirates. Their central idea is the one you just read: whoever competes inside a category fights for a slice, whoever creates the category keeps almost everything. And the most practical lesson in the book is that a category isn't announced, it's taught. That's why you'll see us teaching the market, term by term, until it becomes common sense.