# Challenges in Managing AI Agent Swarms As the capabilities of AI agents are augmenting and we are coming into the agentic swarms, AI agent decoding era, the challenge becomes more around being able to follow the rhythm of agents and being able not to be a blocking factor in the development using AI agents. The challenge there is the decision-making power that is required. So when you are running one agent, then you have some time to think about what's coming next and you stay in context and in flow. But when you are running 2, 3, 5, 10 or more agents, so basically an agent swarm, then the challenge becomes being able to keep up with the rhythm and keep up with the decision-making that's required to give instructions, relevant instructions and feedback to those agents. Actually, this is mentally taxing because it means you have to switch contexts every few minutes from one agent to the next, from one swarm even to the next. As agents' capabilities and speed and performance increase, the tax is only going to increase, and the mental overhead that comes along with agent decoding is just going to increase. So the focus now should be on finding ways to cope with that, possibly with the help of agents. That's yet to be seen, but I think that over time we have to find ways to not be the blocking factors because we are certainly going to be the blocking factors for building with AI agent swarms. I compare this with the ability to manage a huge army in a StarCraft game because you have a base or multiple bases, and you have to give orders to your troops and your groups of troops. You have to perform an attack on one end of the map and construct new things or new armies on another side. You benefit there from the fog of war, but ultimately it's just a strategic game and a game of finding ways to manage everything at once without losing sight of what's happening in the different areas of the game, without losing sight of what the other players are currently doing or planning. So understanding the game, you are playing with your own game with all your different units and groups of units and bases, and following up with what everyone else is doing or planning and adapting your strategy based on that, and also reacting to everything that's happening around. The same is true with agent decoding. Just like the Zerg, who are coming with hundreds or millions of whatever of units, the AI agents swarms are just getting bigger and bigger, and over time this will only increase. --- ## References - Audio I recorded when thinking about this: https://dsebastien.voicenotes.com/challenges-in-managing-ai-agent-swarms ## Related - [[AI Agents]] - [[Cognitive Load]] - [[Decision Fatigue]] - [[Why Humans Are the Bottleneck in AI Agent Swarms and What We Can Do About It (Article)]]