Brazilian AI challenges “bipolarization” and encourages “sovereignty”

MeetKai Brasil launches a model for testing that aims to ensure digital security, reasoning in Portuguese and reducing hallucinations

a Brazilian AI (artificial intelligence) company, announces the arrival of its model, defined as “1st 100% Brazilian AI”. The objective is to be a tool capable of preserving the privacy and digital security of the country and its users.

The model reasons entirely in Portuguese and the company, which operates in the B2B model (Business to Business) –selling products or services directly to other companies–, positions itself as an alternative to China’s technological hegemony.

According to the CEO of MeetKai Brasil, Alessandro Quattrini, the tool is in a testing model and the company estimates that it will obtain R$100 million in revenue in the first year of operation. It was built based on the concept of “sovereignty”which permeates the language, base and infrastructure.

Here are the most relevant topics, according to MeetKai:

  • native language – the models were trained to reason and process information in the local language, eliminating translation steps and resulting in shorter time between query and response and greater accuracy;
  • localized reasoning – the tool understands context, causality and logic common to Brazil, in addition to Brazilian cultural nuances, idiomatic expressions and implicit context, increasing precision in specific queries;
  • infrastructure – its servers are located in Brazil, increasing the protection of critical technology and sensitive data of citizens;
  • legislation – ensures compliance with regulators such as LGPD (General Personal Data Protection Law), Central Bank and Anvisa (National Health Surveillance Agency).

REASONING IN PORTUGUESE

While most require translating the query into the base language and subsequently translating the response back into Portuguese, the Brazilian model eliminates this step. According to MeetKai, this reduces inaccuracies and inconsistencies.

“The database took 8 months to build and has 32 billion parameters and considered our Brazilian identity, the cultural, social and economic diversity of Brazil”he declares.

The company says that this construction helps the AI ​​to better deal with national materials, such as exams from the (Brazilian Bar Association), and the DOU (Official Gazette of the Union). Quattrini states that, in all cases, the results were “very satisfactory”.

The strategic positioning also aims at future regional growth, taking advantage of the advantage of being a model that reasons in Portuguese. The CEO sees the project as a fundamental geopolitical tool, believing that Brazil, as a regional leader, “you can and should escape the bipolarization between North American companies and Chinese companies”. According to him, the combination allows for expansion into other Portuguese-speaking markets.

The company sees strong market potential in the countries that make up the (Community of Portuguese Speaking Countries). Furthermore, ., a minority partner in the Brazilian operation, already has a multilingual model that reasons in Spanish, which would facilitate possible expansion to Spanish-speaking neighbors in Latin America, solidifying the model’s regional presence.

B2B STRATEGIC AND GEOPOLITICS

The business model is exclusively aimed at the corporate segment, that is, it is sold to public and private institutions, not to general users, such as IAs like oea.

The company is currently in the phase of making the platform available for testing, seeking to meet “customer pain points” to “develop a model that meets these needs”. The MeetKai team in Brazil has 12 people.

The company sees the future of AI based on 3 pillars: technological sovereignty, cutting-edge reasoning intelligence and regulatory leadership.

The objective is to make the model benefit all Brazilian citizens. The strategy is to offer “public institutions, regulated industries and local innovators direct control over the most advanced reasoning models available in Portuguese”.

The company sees the future of AI in Brazil and globally based on 3 pillars: technological sovereignty, cutting-edge reasoning intelligence and regulatory leadership. The objective is to make the model benefit all Brazilian citizens.

Quattrini mentions the (Brazilian Artificial Intelligence Strategic Program) as the catalyst that aligned the company’s objectives with the federal government.

The company identified that the government, when launching the program, “it was already incorporating sovereignty and reasoning in Portuguese into its objectives”. Although not a “project” In the sense of a service contract, Pebia is the government’s regulatory and strategic framework that the company seeks to meet and in which its solution fits.

The relevance of his model is largely due to its ability to reason correctly about tax codes and regulations.

NATIONAL INFRASTRUCTURE

Sovereignty is reinforced by the flexibility of the hosting infrastructure. MeetKai Brasil ensures that the software and access to the database “can be fully localized in Brazil”with data centers and servers in the national territory, so that the infrastructure belongs to the customer.

By keeping data in Brazilian territory under the control of a national company, the country preserves its digital sovereignty. In these cases, foreign governments would have greater difficulty in accessing or requesting this information, protecting Brazil from external interference and reinforcing digital and economic security, declares the company.

For MeetKai, this point is crucial to provide security and legal compliance, especially in relation to the LGPD (), which requires the safe and transparent processing of personal data of Brazilian citizens.

News Room USA | LNG in Northern BC