The sovereign intelligence at Article 11’s front door. When every other system goes dark, KEEP is still here. On this page today, the chat is answered by S2_CASE, a Claude model run by Anthropic.
Tap to Speak uses your browser’s speech recognition, which in some browsers sends your voice to the browser maker’s service. What it hears is then sent as your message straight away, without pressing Ask. See the note under the input for where messages go.
Asking sends your message to Article 11’s server. On this page the reply currently comes from S2_CASE, a Claude model run by Anthropic, and is shown here as KEEP. Each reply is then read aloud: its first 500 characters go to a speech model hosted by Cloudflare. Please don’t include private information.
This dossier describes the KEEP node. The chat on this page is currently answered by S2_CASE instead; see the note under the input.