Rémi Louf describes building unattended agentsAn unattended AI agent performs background work without a person supervising every execution step. for market research, voice-note processing and daily briefings. Simple schedules worked initially, but duplicate runs, lost notes and untracked prompt changes exposed ordinary distributed-systems problemsA distributed system is a group of networked components that coordinate to provide one service despite separate state and partial failures. that a prompt alone could not solve.
Rémi Louf replaces a fixed workflow graph with agents that subscribe to typed eventsAn event-driven AI agent reacts to typed events instead of following only one fixed workflow or schedule.. An append-only logAn immutable data store preserves prior records and represents changes by appending new entries instead of rewriting history. preserves every step and causal link, while queues, retries and structured outputs make failures observable and prevent invalid actions from crossing system boundaries.
A content-addressed prompt storeContent-addressable storage identifies stored data by a hash of its content instead of only by a mutable location or filename. records the exact system prompt, skills, tools and user message that reached each model. That representation supports diffs, replay with different models, compaction, cache management and auditing, while keeping the agent definitions simple enough for non-engineers to contribute.
Watch on YouTube
