arOS · Marketing strategy · 2025 to 2026
From 5K to 32K+ users
The platform grew from about 5K to 32K+ users while I led its marketing.
arOS is an AI marketing SaaS. I joined to write prompts and ended up leading its marketing strategy.
Fact sheet
- Role: Marketing Strategist and Prompt Engineer
- Period: Mar 2025 to Sep 2026
- Team: founders, developers and a sales team; I worked closely with co-founder Bruno Picinini
- Scope: positioning, campaigns, launches, landing pages, ads, onboarding and upgrade flow
- Key result: platform grew from about 5K to 32K+ users
Context
arOS turns a company's "DNA" (brand, audience, offer and voice) into content and campaigns through AI agents. When I arrived, it had about 5,000 users. I came in for prompts and agents. My marketing background quickly pulled me into strategy, content and product decisions.
The problem
Marketers are a hard audience to sell AI marketing to. They have seen every promise. We needed to find the segments that would get real value, reach them with proof instead of hype, and make buying and upgrading simple for a product priced at about R$100 a month.
What I did
- Found a segment and tested it fast. Many users were agency owners. We brought in an influencer who runs an agency and already talks about AI tools, and I structured the launch: content, ad messaging, landing page, email sequence, the live lesson outline and the offer.
- Listened to what the segment needed. Agencies asked for multiple seats and team invites. That was not the product's direction, so we scaled back acquisition in that segment even though the media returns were good.
- Moved from sales calls to self-serve. About ten salespeople were working a R$100 a month plan. I adapted the offer, the in-product journey and the upgrade flow so customers could buy on the site or upgrade inside the product, with two formats: a landing page and a landing page with a VSL.
- Made landing page tests cheap. Duplicating pages in a builder slowed us down. I set up a workflow with pages in code on GitHub, a purchased template and AI to produce copy and layout variations. With PostHog reading behavior, we shipped new variations about every one or two days.
- Fixed onboarding with data and conversations. PostHog behavior plus talks with users showed where people dropped off and why some asked for refunds. That fed an AI-assisted onboarding review.
- Created a lot of ads. Ad creative and messaging connected to the same strategy.
Key decisions
| Evidence | Decision | Trade-off |
|---|---|---|
| Agency owners were a large share of users | Test a launch aimed at them with a trusted agency owner | Spend and attention on one segment |
| Agencies needed seats and invites the product would not build | Reduce acquisition in that segment | Gave up a channel that was returning on media |
| Ten salespeople on a R$100 ticket | Move to self-serve purchase and in-product upgrade | Less human touch in the sale |
| Page duplication was slowing experiments | Pages in code, AI variations, PostHog analysis | Required more technical setup |
Results
The platform grew from about 5K to 32K+ users while I led its marketing strategy. This was a team effort: founders, developers and sales all moved the number.
What I learned
- Good media returns don't mean product fit. A segment can pay back and still be the wrong one.
- Pricing shapes the sales motion. A low ticket needs a product that sells itself.
- Fast, cheap experiments beat long debates about a page.
On screen
How it looked in the product.
Landing page tests
A few of the dozens of page variations we tested, from the main sales page to an English version and a quiz funnel.
Onboarding in three steps
Links, Instagram, then the DNA. Each step gives the AI more of the business before the first request.
The help center
I built it end to end: the infrastructure, the copy and the tutorial videos, with guides for arOS 1.0 and 2.0. help.aros.com.br