Founded by Brazilians, the autonomous agent platform Tess AI is moving to Silicon Valley. The move comes on the heels of a US$5 million raised by the startup this month with global investors Hi Ventures, DYDX Capital and Honeystone. Rival to companies like Manus and Cursor, Tess wants to take to the world the idea that AI can work in favor of professionals – and not the opposite.
“When Tess joins a company, in the first six months, four to five pieces of software are canceled. It is the software that is fired, not the employees. The agents’ number 1 enemy are the SaaS, not the workers”, jokes Ricardo Barros, co-founder and CEO of Tess AI, in an interview with Startups.
The move to San Francisco is scheduled for April. The company, which operates with around 30 remote employees, will be based in Silicon Valley to lead its international expansion. Today, 80% to 85% of the customer base is still Brazilian, but Tess already serves companies in 25 countries, including the French Publicis Groupe, the Canadian Maple Bear and the Chinese State Grid.
The latest round of investment brought together names with a relevant history in the sector and strategic for Tess’s new phase. Mexican management firm Hi Ventures is co-led by Federico Antoni, an early-stage investor in Cornershop before its acquisition by Uber. DYDX Capital has as partner Ryan Nichols, former CPO of Salesforce Service Cloud. Honeystone was co-founded by the dean of the Stanford Graduate School of Business, Sarah Soule, alongside professors Jonathan Levav and Yossi Feinberg.
For Rica Barros, the composition of investors is no coincidence. “We validated the thesis technically and academically in this round. The arrival of the former Salesforce executive was very notable because we are living in the SaaSpocalypse era”, he points out. The term describes the ongoing downward movement in the shares of traditional software companies as companies begin to seek AI tools for their day-to-day operations.
Currently, however, most AI companies aimed at corporations have a per-user charging model, which ends up making implementation more expensive – and reducing the return on investment in solutions that would bring more efficiency to the company. Tess positions itself as an alternative to this model. Instead of charging per user, the platform charges per task performed. According to the company, this represents savings of up to 68% compared to ChatGPT Business and up to 90% compared to ChatGPT Enterprise.
Continues after advertising
The proposal also eliminates adoption friction: any employee can create their own agents and share them within the company, without IT approval or extra licensing.
“Vibe working”
Tess’s vision is that, in the future, each employee will have their own team of agents, that is, virtual assistants. “In practice, Tess is only successful if someone within the company is successful. This generates a viral effect, in which one person starts to implement it and it works, and then others start using it. In three months, all departments are using it”, explains Ricardo.
This organic expansion model, which the company calls “vibe working”, has already produced impressive numbers. In one year of operation, more than 16 thousand employees adopted the platform and 2.1 million autonomous tasks were performed. In the last month alone, this number reached 600 thousand tasks – carried out without human intervention.
Continues after advertising
“It’s not just vibe coding, because the professional can have agents helping to carry out tasks that he wouldn’t be able to do alone. They help to unlock skills”, highlights Renato Ferreira, co-founder and COO of Tess AI.
One of the most cited cases by the company is that of the Profarma Group, with around 9 thousand employees. In 90 days, more than 60 employees from areas such as HR, Legal and Finance implemented more than 90 autonomous agents. The result was the creation of a new AI area within the company.
Tess’s platform works as a marketplace with more than 50 thousand different AI agents, which use 268 language models from companies such as OpenAI, Anthropic, Deepseek, Meta, Cohere, Google, among others. The company defines itself as an agentic orchestration platform, not a simple aggregator of AIs.
Continues after advertising
In the GAIA benchmark, an industry reference for evaluating autonomous agents, Tess claims to outperform Manus AI – acquired by Meta for more than US$2 billion – by 10%.
“We launched one of the first agentic orchestration systems in the world. In Tess, when a user makes a request, there is a customized path in which the AIs talk to each other. We also created the idea of AI consensus, in which it is possible to check if the AIs have biases. This way, the user can see what the consensus is between Chinese and American AIs, for example, and compare these views”, explains Ricardo.
The company’s goal for 2026 is to reach US$10 million in revenue, a growth of just over 3x. The expectation is that traction will come from international expansion, but also from organic growth within companies where Tess is already used.
Continues after advertising
For the founders, the move to Silicon Valley is part of a new moment for the startup market. “In the past, the company had to prove itself first locally, in Latin America, for example, before thinking about becoming global. Not now. AI that doesn’t think globally doesn’t exist”, points out Ricardo.
Content produced by .
