Large language model teams face a critical choice: grow by adding more members or learn from past experiences? New research suggests that simply hiring more agents isn't always the answer, especially when budgets are tight. Researchers have developed LLMA-Mem, a framework that teaches multi-agent systems to store and reuse knowledge over time, effectively turning their history into a competitive advantage. The study reveals a surprising twist in how these teams perform: expanding team size does not guarantee better results on long tasks. In fact, smaller groups equipped with smart memory often outperform larger ones because they can efficiently recycle past solutions rather than wasting resources on redundant work. This discovery shifts the focus from brute-force scaling to smarter design, proving that a well-structured memory system is the key to building efficient, high-performing AI teams that get better without necessarily getting bigger. Source: Scaling Teams or Scaling Time? Memory Enabled Lifelong Learning in LLM Multi-Agent Systems by Shanglin Wu, Yuyang Luo, Yueqing Liang, Kaiwen Shi, Yanfang Ye et al., https://arxiv.org/abs/2604.03295