DAEDALUS improves agent memory by learning from self-made tasks
DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks
Artificial IntelligenceComputation and LanguageMachine Learning
Summary
Large language model agents often struggle to act well in new places because they don’t remember what they tried before and keep making the same mistakes. The authors present DAEDALUS, a method where two agents work together: one creates tasks to challenge the other, and from failures, useful strategies or rules are learned. This memory helps the agent do better in future tasks without relying on human instructions or prior knowledge. Their tests show DAEDALUS boosts task success noticeably and works even with limited practice.
What this means in practice
- •For automation engineers: Improve tool-using agents in unknown environments by enabling them to learn and remember useful strategies autonomously.
- •For software developers building ai assistants: Deploy memory-augmented agents that adapt to new tasks faster without task-specific training or human guidance.
Authors
Antoine Edy, Max Conti, Victor Xing, Marc-Antoine Allard, Nawfal Benhamdane, Gautier Viaud
Abstract
LLM agents often lack the operational knowledge to act reliably in new environments, as they must discover specific tool behaviors or environment conventions on their own. Without memory of past attempts, they repeat the same mistakes across tasks, leading to more task failures and longer trajectories. To address this, agentic systems typically rely on human-written guidelines or on procedural memory built from training tasks and an oracle verifier, both of which require prior knowledge of the environment. We present DAEDALUS, a method for bootstrapping reusable agent memory from self-generated practice without existing tasks or oracle verifiers. DAEDALUS pairs two agents: an explorer that interacts with the environment to generate challenging yet solvable tasks, and a solver that attempts them. A heuristic is derived from each solver failure and accepted only after the solver repeatedly succeeds with that heuristic in context. These outcomes also provide feedback for the explorer to refine the difficulty of future tasks. Accepted heuristics are then consolidated into a memory bank for test-time use. Across AppWorld, $τ^2$-bench, and AutomationBench, DAEDALUS improves mean success rates by up to 15.9 points and pass^5 by up to 2.2x over a no-memory baseline, and is competitive with methods using training tasks, at a lower inference cost than most. We show that performance gains already emerge with a small exploration budget, and that its heuristics also benefit agents from other model families. Our ablations further reveal that solver traces provide the key information needed to derive effective heuristics, while factorizing early discoveries makes exploration more cost-efficient. Beyond memory construction, we find that the tasks generated by DAEDALUS can serve as a proxy for benchmark tasks when ranking models by performance. Code and artifacts: www.github.com/illuin-tech/daedalus.