Serve para situações em que documentos, sistemas de IA e decisão humana se misturam e alguém precisa, mais tarde, de reconstruir exatamente o que aconteceu.
It is for situations where documents, AI systems and human judgment interact and someone later needs to reconstruct exactly what happened.
Junta a evidência essencial do caso num pacote verificável e resistente a alterações silenciosas.
Normaliza exports JSON de OpenAI, Anthropic ou outros sistemas para um formato comum, mantendo explícitos o fornecedor e o hash da fonte original.
Permite contestar uma decisão sem apagar nem reescrever o registo original.
Pode bloquear o estado “pronto para decisão” quando faltam elementos críticos.
Collects the essential case evidence in a package that can be independently checked and resists silent alteration.
Normalizes JSON exports from OpenAI, Anthropic or other systems into one common format while keeping the provider and original-source hash explicit.
Lets a later challenge be recorded without erasing or rewriting the original decision record.
Can block “ready for decision” status when critical evidential elements are still missing.
Integridade criptográfica não significa que o conteúdo seja verdadeiro.
A decisão final humana continua separada da análise produzida pela IA.
O hash prova a correspondência do ficheiro recebido; não certifica, por si só, que o fornecedor emitiu esse ficheiro nem garante conformidade regulatória.
Cryptographic integrity does not mean the underlying content is factually true.
The final human decision remains distinct from the analysis produced by AI.
The hash proves correspondence to the received file; it does not by itself certify who issued that file or guarantee regulatory compliance.
Bancos, seguradoras, auditoria, equipas jurídicas, administração pública, tribunais, investigações internas e outras organizações em que uma decisão apoiada por IA tenha de ser explicável e reconstruível depois do facto.
Contacto: ricardocrespo52@gmail.com
Banks, insurers, audit teams, legal teams, public administration, courts, internal investigations and other organisations where an AI-assisted decision may later need to be explained and reconstructed.
Contact: ricardocrespo52@gmail.com