Brazil's Chamber of Deputies (Câmara dos Deputados) has operated Ulysses, an in-house suite of AI and natural-language-processing tools for legislative work, since 2018. The suite grew out of the Chamber's Digital Strategy (Act 268/2021) and now comprises eight modules, labelled U1 through U8, covering distinct parts of the legislative workflow.
U1 aggregates thematic content for the Chamber's legislative-studies portal. U2 semi-automatically routes citizen and staff requests to the relevant legislative-counsel groups. U3, the most heavily used module, runs semantic analysis to flag bills similar to a new proposal and processes more than 20,000 such requests every year. U4 clusters amendments using NLP, U5 transcribes plenary and committee speech (built on Microsoft Azure speech-to-text, with a migration to OpenAI Whisper planned), and U6 analyses public e-poll comments on bills. U7 and U8 apply biometric facial recognition, respectively to authenticate parliamentary votes through the Infoleg app and to tag legislators automatically in official photographs.
In December 2025 the Chamber announced a new phase of the programme: an internal 'Ulysses Chat' assistant for staff, formal permission for employees to use external tools such as Claude, Gemini and GPT, and a new set of AI governance and security policies to accompany the expansion. No independent evaluation of the modules' accuracy, error rates or legislative impact has been published, and the biometric voting and facial-recognition modules operate without a named external oversight body.
Read the full analysis: https://www.mobiletime.com.br/noticias/15/12/2025/ulysses-camara-nova-fase/
Where this practice's information was retrieved from, and when.