GrowthX AI

A growth team in a box, minus the box.

On a Wednesday morning in San Francisco, a model running inside GrowthX is drafting a piece of writing that will, three weeks from now, be quietly cited by ChatGPT when a procurement lead at a Fortune 500 asks which fraud platform to evaluate. A human editor at GrowthX will touch the file twice. A founder at Abnormal Security will never see the cursor blink. This is the entire pitch.

GrowthX AI is what happens when you stop pretending an agency is a software company and start building it as one. The company sells a single product with two faces - dynamic AI workflows on the back, forward-deployed humans on the front - and rents the whole apparatus to brands that want to keep showing up when buyers stop typing URLs. Reddit pays for it. So do Webflow, Ramp, Superhuman, Bolt, Strapi and roughly thirty-five other names you would recognize. As of May 2025, Madrona pays for it too, in the form of a $12 million Series A.

It is, at minimum, a strange company. It calls itself a platform but sounds like a consultancy. It runs profitably and then raises money anyway. It writes blog posts about buying habits in 2026 while quietly indexing the citation graph of large language models. Two years ago, "AI content" was a slur. GrowthX has somehow made it a line item.

The problem they saw before anyone else admitted it.

The conventional wisdom in growth marketing, for roughly twenty years, has been: rank on Google, capture the click, route the click into a funnel, optimize the funnel. The funnel is dead, or at least gravely embarrassed. Buyers ask Perplexity. Buyers ask Claude. Buyers ask Gemini. The funnel now begins inside a model that may never link out.

The traditional response from agencies has been to write more blog posts faster. The traditional response from AI tools has been to write more blog posts even faster, and worse. GrowthX's read was different and slightly heretical: nobody actually needs more blog posts. They need to be the answer the model returns. That is a different sport.

Once you accept that framing, everything downstream shifts. Keyword research becomes prompt research. Ranking becomes citation share. Editorial calendars become opinionated knowledge graphs. The agency model - hire a junior writer, ship a brief, iterate over six weeks - breaks completely. You need software-shaped operations and writers who think like analysts. There were, until recently, almost zero companies set up to do both.

The founders' bet.

Marcel Santilli has spent fifteen years inside marketing teams at companies that ended up large: Scale AI, HashiCorp, ServiceTitan, Deepgram. The arc is consistent. He arrives, builds a growth engine, the company doubles, he leaves. Three of those companies became unicorns. He likes to say he is responsible, in a small and shared way, for north of $1 billion in combined ARR. The number is hard to audit and easy to believe.

Daniel Lopes ran engineering at Canopy, the 37signals spin-off, which is the kind of credential that translates poorly into pitch decks but very well into product. Together, in 2024, they made a calculated wager: that the next decade of marketing software would look less like SaaS and more like an opinionated services firm with software in its bones. They called it services-as-software, an unlovely phrase that nevertheless describes the entire roadmap.

The bet was not just product-shaped, it was business-shaped. They charged from day one. They ran workshops. They sold pilot contracts. By the time they spoke to Madrona, they were running profitably, growing fast, and selling a thing that the venture community was still arguing about over dinner. That is an unusual posture for an AI startup in 2025, when most companies in the category lead with vibes and follow with losses.

The product, briefly described.

What GrowthX sells looks, from outside, like a managed service. From inside, it looks like a CI/CD pipeline for content. A customer signs a retainer (the full engine starts at $18,000 a month, which is closer to agency pricing than SaaS pricing, on purpose). GrowthX builds a custom set of AI workflows tuned to the customer's category, voice and citation goals. Forward-deployed editors and strategists - the "expert humans" in every deck - run the system in production, refine its prompts, and ship work daily.

The intended outputs are unfashionably specific: rankings, citations, organic traffic and pipeline. Vanity metrics are conspicuously absent from their public posts, which in this category is its own kind of marketing.

The proof, in customers and numbers.

The case for GrowthX is, mostly, the customer logo wall. It is hard to sell content services to Reddit, which has a slightly above-average opinion of itself when it comes to internet writing. It is also hard to sell content services to Webflow, whose brand team is famously discerning, or to Superhuman, whose entire pitch is taste. GrowthX has somehow signed all three.

The mission, plainly.

Strip the language back and the mission is small enough to fit on a napkin: turn content into compounding organic growth, and do it by making your brand the answer wherever buyers search - and wherever AI answers for them. The first half sounds like every content agency since 2008. The second half is the bet.

It is also the part that is most exposed. AI search behavior is changing every quarter. The economic value of a citation inside GPT or Gemini is still being negotiated, often by lawyers. GrowthX has built a business on the assumption that the answers people get from models will become the new front page of the internet. That assumption could turn out to be early, or expensive, or both. So far it is looking correct.

Why it matters tomorrow.

If GrowthX is right about one thing in particular, every B2B marketing org will be rebuilt around AI workflows inside three years - not because executives suddenly love AI, but because the unit economics of writing a thousand strong pages a month with a team of fifty no longer makes sense. The company that learns to ship that work first, cleanly, with verifiable quality, captures the category.

There are reasons to be skeptical. The agency category is brutal, low-margin and faintly cursed. Software companies that try to do services tend to drown in human costs. Services companies that try to do software tend to drown in scope. GrowthX has chosen to do both at once, on purpose, and is somehow still buoyant.

Back to the Wednesday morning in San Francisco. The draft that the model wrote at dawn ships at 4pm. Three weeks later, ChatGPT cites it - not because anyone gamed the system, but because the system was built to be cite-worthy. A procurement lead at a Fortune 500 evaluates the company that ran the draft. A demo gets booked. A contract gets signed. None of it goes through Google. None of it goes through a click. This is the thing GrowthX is selling. It is, increasingly, the only thing left worth buying.

Links, archives, video.