# Danilo Parmegiani > Danilo Parmegiani is a product builder based in São Paulo, Brazil, and the co-founder of Groselia, an AI native social media agency, where he leads product, design and technology. He works across design and engineering: he designed Groselia's brand and product and built the platform it runs on with AI coding agents. ## Story I'm Danilo, a product builder and Groselia co-founder based in São Paulo, with hands-on experience designing, building and shipping AI products for real customers. I combine product and brand design, full-stack development and customer discovery with data analysis at Brex and enterprise product prototyping at IBM. My first job, in customer service at Sabesp, paid for my tuition. At IBM Client Engineering I ran discovery with enterprise clients across Brazil and Latin America and turned what we heard into watsonx prototypes and working demos in 4 to 8 week engagements. At Brex I learned how money actually moves: I dug into payment operations with SQL and Python, turned exceptions into requirements for Product and Engineering, and built a GPT-4 Slack summarizer that was showcased at an internal All-Hands. Groselia is where it all comes together: a two-person team, no playbook, and a platform I designed and built from the database up. ## Experience - Co-founder, Groselia (Apr 2026 to now). Co-founded Groselia and lead product, design and technology. We're building an AI-native marketing service for technology companies, and my work connects customer needs and product priorities with hands-on AI development. - Shape the marketing engineer role, combining marketing strategy, creative judgment and AI deployment to run client delivery through repeatable workflows. - Translate business goals into working systems through client discovery, process redesign, prototyping, API integrations and implementation. - Build conversational AI agents with tool use, web research, streaming responses and persistent brand knowledge. - Build across TypeScript and React, Supabase (Postgres, Auth, Storage, Edge Functions), Vercel and Trigger.dev background jobs. - Integrate Stripe subscriptions and AI cost tracking, connecting product usage with billing and account access. - Turn websites, documents and social media data into structured context that reflects each client's brand, strategy and goals. - Raised an initial pre-seed round with the founding team from Brazilian venture investors. - Banking Operations Analyst, Rotational, Brex (Jan 2025 to Mar 2026). Worked across Banking Operations and Payables, translating operational problems into data analyses, automation opportunities and requirements for Product and Engineering. - Analyzed escalations with SQL, Snowflake, Hex and Python; built taxonomies and found the recurring root causes behind them. - Partnered with Product, Engineering, CX and Risk to turn payment exceptions into requirements, and validated proposed backend logic against transaction data. - Built a Slack summarization proof of concept with GPT-4 through Brex's internal LLM gateway, surfacing decisions and next steps. Showcased at an internal All-Hands. - Intern, Operations, Brex (Feb 2024 to Jan 2025). Focused on improving ACH and check-review workflows within Banking Operations, combining operational data analysis with process improvement and Engineering collaboration. - Identified manual steps and bottlenecks in payment review workflows. - Analyzed transaction data to validate proposed solutions and investigate operational exceptions. - Partnered with Engineering on backend logic improvements, translating operational needs into technical requirements. - Helped reduce manual workload while staying aligned with compliance and operational standards. - Intern, Client Engineering, IBM (Mar 2022 to Feb 2024). Worked in IBM Client Engineering, an enterprise presales team supporting AI solution discovery and prototyping for selected clients in Brazil and Latin America, with a focus on IBM watsonx. - Supported 4–8 week MVP engagements, translating business challenges and user needs into prototypes and working demos. - Conducted client interviews and user research to find practical use cases and clarify requirements. - Built application prototypes in Figma with developers, data scientists and designers, connecting user experience with technical delivery. - Created client presentations, solution demos, videos and engagement documentation. - Founded and led InternWays, an internal podcast connecting interns with senior executives. - Intern, Marketing & Content, PHS Brasil (Sep 2020 to Jan 2021). Created B2B technology marketing content for a consultancy, translating technical topics into material for customers and prospects. It gave me firsthand exposure to how repetitive content operations are, a problem I later returned to with Groselia. - Produced content and campaign assets for LinkedIn and Instagram. - Structured email marketing campaigns in RD Station. - Developed e-books and other lead magnets for clients and prospects. - Wrote cybersecurity content for the company's blog and social channels. - Intern, Customer Service, Sabesp (Aug 2018 to Mar 2020). My first job, and it holds a special place for me. I supported customers by phone and chat with questions about water and sewage services, billing, tariffs and installment plans. - Explained service and billing information in accessible language. - Helped customers navigate service issues, including water and sewage leaks. - Evaluated fellow interns against service standards and reported findings to management. ## Education - Bachelor's degree, Business Administration, Link School of Business (Feb 2021 to Dec 2024). ## Certificates - Business Strategy, Tel Aviv University (May 2023) - Entrepreneurship Strategy, The Wharton School (Dec 2022) - Enterprise Design Thinking Practitioner, IBM (Nov 2022) - Excellence in Business and Leadership, Disney Institute (May 2022) - Sharpen Your Communication Skills, Stanford Graduate School of Business (Oct 2021) ## Languages - Portuguese: Native - English: Advanced ## Case studies ### Groselia (2025–now) https://daniloparmegiani.com/en/groselia Groselia runs social media for tech startups first, and for companies of other industries and sizes too: strategy, design, publishing, comment replies and reporting, on every channel, from one platform. I started Groselia with Pedro as a team of two. He runs marketing, sales and operations; I designed the brand and the product and built the platform it runs on, from the database to the last animation. While it was self-serve, more than 200 people signed up to use it, and today it runs our customers' content and social media. In June 2026 we raised a R$100k pre-seed from Link Vector, the accelerator of Link School of Business, where I graduated, and two months after the pivot to a service, Groselia brings in more than R$15,000 a month. - Role: Co-founder, Engineering, Product and brand design. - Team: Danilo Parmegiani (Co-founder), Pedro Critchi (Co-founder). - Tools: Codex, Claude Code, Figma, Adobe tools. Stack: React, Next.js, TypeScript, Tailwind, Supabase, Frontier LLMs, Trigger.dev, Stripe, Vercel. - By the numbers: R$180k+ annual revenue run rate, 200+ accounts created, 350 brands set up, 9,700+ commits, 520+ pull requests, 47 edge functions. - Links: Website https://www.groselia.com/en, Open the app https://app.groselia.com. The problem: Building stopped being the bottleneck. Being seen became it. AI collapsed the cost of making software. A founder with no engineering background can ship a product in a weekend now, and so can everybody else. What didn't get cheaper was getting anyone to notice it. Distribution is a continuous operation, not a task: posting every day, on eight networks that each want a different format, answering every comment, and reading numbers that each app reports differently. Startups pay for that in one of three ways. They hire a team they can't afford yet, they stitch together a dozen tools that each solve one slice, or they post nothing and pay with growth. Groselia exists to be the fourth option. How it got here: - Late 2025 – Feb 2026, Argibee: An AI brand-design platform. It started as a prototype at the end of 2025 and became a real company in February 2026. It didn't find its market, but its brand-kit engine became the spine of everything after it. - Apr 2026, Motion video: Groselia started as Apple-style launch videos made from After Effects templates. I validated the whole pipeline by hand, and then saw it couldn't scale to the quality bar. - Jun 2026, Pre-seed from Link Vector: We joined Link School of Business' Startup School, a 100-day program for founders filmed as a reality show (watch the season (https://www.youtube.com/playlist?list=PLpkuohtdvWTBvMolsqSTtEJznmDc_8riv)). It closed with a demo day where I pitched Groselia on stage, and with our R$100k pre-seed from Link Vector, the accelerator of Link School of Business, where I graduated. - Aug 2026, Lovable for marketing: One conversation instead of a form: describe the campaign and Groselia plans, writes and publishes it. A chat agent with tools, not a roster of bots. - Sep 2026, A social media manager: Customers didn't want AI to make their content. They wanted their own photos posted well and their comments answered. We cut generation and went deep on running the channels. - Now, Service as a software: We sell the work, not the tool. Our team runs each customer's social media on the platform, which makes one person able to run what used to take an agency. How it was built: I built Groselia with AI agents as my engineering team. Codex and Claude Code wrote the code, each change on its own branch, while I designed it, reviewed every pull request and decided what shipped. And I spoke most of my prompts instead of typing them, with Wispr Flow. A good prompt is a long, detailed brief, and talking let me give one as fast as I could think it. Takeaways: - Taste is the moat: When AI makes building cheap, everyone can ship, and most of it looks the same. What's still scarce is judgment: knowing what to leave out, how a screen should feel, when an animation is one too many. For a team of two that can't outspend anyone, taste is what makes Groselia look and work like something much bigger, and it's the one part I'd never hand to a model. - Sell the outcome, not the tool: Nobody wanted to operate marketing software, however good. They wanted to be seen. Selling the work instead of the app made the product matter more, not less: it became the engine we run every day, so every rough edge got fixed. - Context beats cleverness: The agent got better when it knew more, not when the prompt got smarter. Giving it the live calendar, the memory and the open plans every turn, and making it ask before it acts, did more than any model upgrade. - AI changes the size of the team: With agents writing most of the code in parallel, my job moved to deciding what to build, reviewing every change and keeping the bar high. One person can now run what used to take a team, as long as the person stays the editor. ### Coisa de Motos (2025) https://daniloparmegiani.com/en/coisa Coisa de Motos sells motorcycle gear, apparel and parts online, to a community that started around a YouTube channel in 2020. We built the store on Shopify, and I designed it along with the brand and the content that brings people to it. The channel already had an audience of riders, the motonautas, but no identity that could carry a store. The brand had to work everywhere a rider looks: a helmet sticker, a cap, a YouTube thumbnail, a story, and the header of an e-commerce site with hundreds of parts. A few months after launch we closed it: each partner wanted to take the company somewhere different, and we all had other opportunities in our careers. - Role: E-commerce development and design, Brand identity, Content and motion. - Team: Daniel Critchi (CEO and co-founder), Danilo Parmegiani (Co-founder), Pedro Critchi (Co-founder). - Tools: Shopify, Illustrator, Photoshop, After Effects. - Deliverables: Shopify store, Mobile shopping, Promotions, Blog, Identity, Mascot, Ads and thumbnails. The challenge: Coisa de Motos was a name, a voice and a community before it was anything you could buy. Turning it into a store meant keeping that voice, loud, funny and very much from the road, while looking serious enough for someone to buy a R$2,000 jacket from it. It also had to stretch. The same identity would sit on merch, on a site full of brake discs and visors, and on a weekly rhythm of stories and thumbnails, made fast and always recognisable. The store: We built the whole store on Shopify and used as much of the platform as we could instead of building around it: a theme customised to the brand, the catalogue with its collections and variants, search and filters, checkout and payments, discount codes like MOTONAUTA, the blog, and apps for the rest. Leaning on the platform kept a two-person team selling hundreds of parts without running servers, and left our time for what makes the store different: how it looks, how it's organised and what it says. Customising a Shopify theme means writing Liquid, Shopify's own template language, mixed with HTML, CSS and JavaScript. I wrote ours with ChatGPT. I'd describe the change I wanted, paste the code it gave me into Shopify's theme editor, and see what happened. When something broke, I copied the code back out of Shopify and sent it to ChatGPT with the bug, until it worked. AI coding was still early, and none of it lived in the theme editor I worked in. It was copy and paste between two tabs, but it was already how I got things built. Takeaways: - Knowing when to stop: We closed Coisa de Motos a few months after the store went live. Each of us wanted to take the company in a different direction, and each of us had other opportunities in our careers, so we stopped rather than pull it three ways. The brand, the store and everything we made for it are what's left, and what this page is about. - A community is a brief: The best decisions came from the riders themselves: their word for themselves, motonauta, became the coupon, the greeting and half the content. - Design for the fastest use: Most of the brand is used in a hurry, a story before lunch, a thumbnail before upload. Templates that are hard to get wrong did more for consistency than any guideline. - One loud colour is enough: Keeping orange for what matters, the symbol, the button, the offer, made it work harder than a bigger palette would have. ## Contact - Email: parmegianidan@gmail.com - LinkedIn: https://www.linkedin.com/in/daniloparmegiani/ - GitHub: https://github.com/danilo-parmegiani - Instagram: https://www.instagram.com/dan_parmegiani/ (@dan_parmegiani) - Contact page: https://daniloparmegiani.com/en/contact ## Pages - [Home](https://daniloparmegiani.com/en): Who Danilo is, where he has worked, the tools he uses, and live pieces of the Groselia app. - [Projects](https://daniloparmegiani.com/en/projects): Every case study, with a short summary of each. - [Groselia case study](https://daniloparmegiani.com/en/groselia): How Groselia got here, its brand, the product and its AI agent, the website, the engineering, the technical challenges and the takeaways. - [Coisa de Motos case study](https://daniloparmegiani.com/en/coisa): A motorcycle community turned into a brand and a Shopify store: the e-commerce, the identity, the content and the takeaways. - [Resume](https://daniloparmegiani.com/en/resume): Experience, education, certificates and languages. - [Contact](https://daniloparmegiani.com/en/contact): Email, Instagram, LinkedIn and GitHub.