São Paulo's disability-rights secretariat embedded Hand Talk's AI avatar Hugo across all 32 subprefecture and 26 secretariat portals: 40 million Portuguese-to-Libras translations logged by 2023, over 10 million words by 2025, serving 145,000 residents with hearing difficulty.
40M
Cumulative translations delivered (to Aug 2023)
10M+
Words translated milestone (to Aug 2025)
Details
Maturity
Established
Promoter
Secretaria Municipal da Pessoa com Deficiência (SMPED), Prefeitura de São Paulo, with Hand Talk
Period
2023 – ongoing
Keywords
e-government, disability inclusion, accessibility, sign language AI, digital public services
Context
Sao Paulo's Secretaria Municipal da Pessoa com Deficiencia (SMPED) has embedded an AI-driven Portuguese-to-Libras translation plugin from Hand Talk across the city's entire official web presence -- the central municipal portal, all 26 secretariats and all 32 subprefectures. A clickable icon triggers an avatar, Hugo, which signs page text in real time using AI rather than pre-recorded human video.
Results
In August 2023 the city reported 40 million cumulative Portuguese-to-Libras translations delivered across its portals. In August 2025 it reported a further milestone of over 10 million words translated, citing 2022 Census data showing more than 145,000 Sao Paulo residents with permanent hearing difficulty.
Conclusions
No independent, third-party evaluation of translation accuracy, linguistic fidelity or user satisfaction was found; AI-generated sign avatars have drawn documented concerns elsewhere in Brazil's Deaf community and linguistics research about grammatical fidelity compared with human interpreters.
Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
Municipal portal content and accessibility teams integrating the plugin
Conditions for success
Legal mandate under Brazil's 2015 Lei Brasileira de Inclusao web-accessibility law
Off-the-shelf vendor product embedded across all municipal digital touchpoints
Common failure modes
No independent evaluation of translation accuracy or linguistic fidelity
Deaf-community and linguistics concerns about grammatical fidelity vs human interpreters
Commonly funded by
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.