I’m one of the builders of LOLM, a hybrid Transformer–SSM language model and agent system from Qira. The central idea is that an agent should do more than generate text and call tools. LOLM exposes a controller that can decide when to continue, retrieve evidence, verify, branch, or finalize, and it produces a receipt showing what happened during the run. Available now: - Live agent demo - CLI for questions, code tasks, and small HTML builds - Isolated coding sandbox - Local/self-hosting path...