About Pixelwolf
Building technology that actually works, since 2015.
We started as a digital product studio, were acquired, spent three years inside an edtech operating at scale, and then took the brand back. Today we're an AI consultancy and software house for business operations, working with companies in Brazil and abroad.
Our story
From the first screen to AI inside the operation.
Pixelwolf was founded in Curitiba in 2015 as a digital product studio obsessed with user experience. Understanding the problem came before any screen, so much so that discovery was offered free of charge. That's how we came to run all of Jovem Nerd's technology, from website to app, and went on to build for XP, EBANX, Hotmart, Ailos and Porta dos Fundos.
AI became part of our work early. In 2018, years before the generative AI boom, we were already working on AI projects alongside IBM and Microsoft. In 2020, Wollo, a product built by Pixelwolf, was acquired by Hotmart. In 2021, Pixelwolf itself, then around 40 people, was acquired by Fluency Academy, an edtech backed by General Atlantic, and became the company's embedded technology team.
During those years we built product, technology and courses at scale, with AI applied to education tools. At the end of 2024 we took the brand back with the founding team and a clearer mission: bring to other companies what we learned making AI work inside a real operation.
Today we work across four areas: process diagnosis, AI agents and automation, custom AI solutions and team training. The work covers areas such as customer service, sales, marketing, legal, finance and the shop floor.
Manifesto
Good technology is the kind that disappears into everyday work.
AI has become a commodity. Bringing real intelligence into a business hasn't. The difference isn't the tool; it's how deeply you understand the operation before automating it. That's why we always start with processes, using an immersion method that looks at how the work actually gets done, and we keep people in charge. Automation takes on the repetitive work; decisions and relationships stay human.
Processes → AI → Results.
Three principles behind every project.
-
The operation comes first.
No recommendation starts from a tool looking for a problem. It starts from a mapped process, with the bottleneck identified and the gain estimated.
-
Tool-agnostic by principle.
We don't resell tools or partner with AI vendors. Every option is presented with pros, cons and cost, and the decision is yours.
-
Builders diagnose better.
The team that designs the plan also builds products, both our own and our clients'. Nothing goes into a roadmap that we don't know how to deliver.
From diagnosis to AI that lasts.
The path changes from company to company, but the order is always the same: understand first, then prioritize, and only then build.
- 01
A 4 to 6 week diagnosis
We interview the board, managers, leaders and operational teams to understand how the work gets done, where the bottlenecks are and what data already exists.
- 02
Prioritization matrix
Each opportunity is weighed by effort, risk and return, separating quick wins from long-term projects. Usage and data governance is designed alongside the plan.
- 03
Implementation in 12-week cycles
Agents, automations and custom solutions go live in short cycles, with deliveries the team is already using along the way.
- 04
Continuous improvement
We measure, adjust and train people so the AI keeps working and improving after delivery.
What we've learned
The four mistakes we see most in AI projects.
Exame (Brazil) · September 28, 2026
Pixelwolf's founder on the most common mistakes companies make when implementing AI
Read the article (in Portuguese) ↗-
Starting with the tool
Putting AI everywhere without knowing what it's meant to solve. Without a defined problem, there's no way to measure return.
-
Rolling out without governance
Handing tools to the team with no rules for usage, data and access. The risk tends to show up before the results.
-
Shutting everything down after the first incident
Blocking all AI use after a scare. The company loses what it has learned, and usage tends to carry on, just off the radar.
-
Trying to build everything in-house
MIT research found that AI projects run with specialized vendors have roughly twice the success rate of those built internally.
The milestones so far.
- 2015
The beginning
Pixelwolf is founded in Curitiba as a digital product studio focused on user experience.
- 2016
Entertainment at scale
All of Jovem Nerd's technology, plus products for Festival de Curitiba.
- 2018
AI and community
First AI projects alongside IBM and Microsoft. Our new Curitiba office becomes a community hub, home to Kotlin CWB and Figma Curitiba.
- 2019
Diversification
Projects for Ailos and Hotmart, in finance and education.
- 2020
Wollo
Wollo, a product built by Pixelwolf, is acquired by Hotmart.
- 2021
Acquisition
Pixelwolf is acquired by Fluency Academy, an edtech backed by General Atlantic, and becomes its technology team.
- 2021 to 2024
AI in education
AI applied to products and digital tools for the education market.
- 2024
Pixelwolf 2.0
At the end of the year, we take the brand back with the founding team and a focus on AI.
- 2026
AI applied to operations
Diagnosis, implementation, custom projects and training, in Brazil and abroad.
Team
A senior team, from strategy to engineering.
People who have built for large companies and know that results come from intelligence, knowledge and process. Each professional coordinates their own AI agents, which shifts people's work towards architecture, review and decision-making. The team is led by Thiago Bisewski, Pixelwolf's founder, who has more than 15 years in technology and teaches applied AI in the Exame | Saint Paul MBA and extension courses in Brazil.
“Writing code used to be almost therapeutic. Today it's a job of review and architecture, and that's where experience and range really make the difference.”
Brands that have built with Pixelwolf.
Ready to talk about your operation?
In a first conversation, we get to know your context and point out where AI creates the most value.