Teia Studio is selected as an ALPHA Startup for the country’s largest technology event and will present the first global framework that audits, measures, and monitors in real time what AIs say about any company
A Brazilian startup will present an unprecedented global platform at Web Summit Rio 2026, held from June 8 to 11, capable of auditing in real time what the main generative artificial intelligences say about brands, institutions, and services. Teia Studio, selected as an ALPHA Startup for the event, will showcase its solution on June 9 at Booth A3-30, competing in the PITCH Competition — Web Summit’s premier startup contest.
The founding team will be available for live demonstrations of the Teia GEO IGO platform and interviews regarding the results of its Pilot Program, which revealed alarming data about the behavior of generative AIs in Brazil.
The Problem Teia Studio Solves
Daily, millions of people use ChatGPT, Gemini, Claude, and Perplexity as their primary source of information regarding companies, hospitals, products, services, and even politicians. However, until now, no one was monitoring what these AIs actually say. And what they say is not always true.
Generative AIs do not work like Google. While the search engine displays a list of links and lets the user decide which to trust, a generative AI delivers a pre-packaged narrative presented as the absolute truth — with no page 2, and no second opinion.
“The GEO market already arranges companies’ digital storefronts to appear in AIs, but until now, no one has been checking in real time what these AIs are actually telling people. That is exactly what Teia GEO IGO measures,” explains José Vásquez, founder and CDAO of Teia Studio.
How the Teia GEO IGO Platform Works
Teia GEO IGO operates 24/7 in weekly audit cycles. The process works as follows:
- Source of Truth: The platform starts with a company’s official website — already indexed by Google — as the original source of truth;
- Simultaneous Queries: Standardized questions are sent simultaneously to four leading AI models: ChatGPT, Gemini, Claude, and Perplexity;
- Data Logging: Each response is fully recorded in a database, including the date, time, and specific model;
- Automated Audit: The system audits whether the brand was mentioned, whether the information is accurate or fabricated (hallucinations), the degree of agreement among models, and behavior variations over time;
- Real Dashboards: The results are consolidated into auditable dashboards and reports featuring real data, rather than simulations.
To make this data comparable and actionable, Teia Studio developed proprietary indicators called KAPIs (Key Algorithmic Performance Indicators), which transform AI perception into auditable metrics.
The methodology was named IGO — Intelligence Governance & Observability, a new discipline that goes beyond GEO (Generative Engine Optimization). While GEO only seeks to optimize how brands appear in AIs, IGO continuously audits and monitors what these AIs effectively say. The methodology is protected by a patent application filed with the INPI (BR 10 2026 001032-4) and is based on a scientific paper published on Zenodo (DOI 10.5281/zenodo.19765674).
What the Data Reveals: Pilot Program Results
Between March 9 and April 7, 2026, Teia Studio conducted its Pilot Program with four real brands from four distinct sectors, auditing four AI models in weekly cycles. In the case study of the public health sector — spanning two cycles and 640 analyzed interactions —, the results were compelling:
- 58 hallucinations were detected, 18 of which were classified as high-severity — meaning the data was completely invented by the AIs;
- A 1,250% increase in hallucinations occurred between the first and second week (jumping from 4 to 54 occurrences);
- Perplexity accounted for 83% of the severe hallucinations, creating fabricated medical registration numbers (CRM), incorrect geographic locations, non-existent websites, fabricated scientific studies, and fake contact information;
- In 1 out of every 3 questions, the four AIs gave contradictory responses to each other regarding the exact same query;
- On average, the AIs only mentioned the brand in 44% of the responses — in the remaining instances, they simply ignored the institution even when the question was directly about it;
- No AI warned the user about the possibility of providing incorrect information.
Based on these real-time diagnoses, the TEIA GEO IGO platform suggests technical actions on the brands’ official domains and social networks so that AIs begin to recognize official content as the primary source of truth.
“With TEIA GEO IGO, companies begin to see, in a structured way, what AIs are saying about them — and how that narrative compares to their competitors. This completely changes strategic decision-making capabilities across marketing, reputation, and commercial operations,” concludes Vásquez.
More information: https://teiageo.com/
Press Relations: André Sales – andre.sales@saleslima.com.br – +55 11 98254 4477
| Contatos da assessoria |
| Tel: +55 11 98254-4477 |
| Email: andre.sales@saleslima.com.br |
Click here to see more news.
