When an AI system issues a life-or-death recommendation, who stays in charge – the algorithm or the clinician? TANGO answers: both!
Developed by a consortium of 8 partners and more than 40 open-source contributors, the TANGO Platform is a modular, Apache-2-licensed Python stack that lets humans and machines share decision authority without sacrificing transparency or efficiency.
TANGO ships three plug-and-play building blocks:
– fairness & ethics core: auditable bias-detection and mitigation pipes that log every rule check;
– explain-for-whom layer: context-aware explanations tuned to different audiences (experts, citizens or regulators);
– hybrid-learning API: a REST interface (inspired by the OpenAI pattern) that injects human feedback to enhance models – live.
These blocks already power four pilots in high-impact case studies (maternal-care triage, intra-operative guidance for surgeons during the treatment of abdominal aortic aneurysms, credit allocation by financial service companies and scenario building for policy makers) showing how an ethical AI loop can cut diagnostic delay or flag discriminatory lending in live working conditions.
In Bolzano we will walk through the GitLab repos, wire up an explainer to a sample classifier and stream a live counterfactual explanation – all in under eight minutes – then share roadmap tasks where the community can jump in.
If you are hunting for a pragmatic way to weave EU AI-Act compliance and genuine value alignment into your own codebase, TANGO is yours to fork, probe and improve. Join us and help keep humans in the decision loop – by design, and in production.
