Added agent architectures

mongodb-developer · Feb 15, 2025 · a1e8c0b · a1e8c0b
1 parent 3d7a349
commit a1e8c0b
Show file tree

Hide file tree

Showing 6 changed files with 33 additions and 5 deletions.
diff --git a/docs/10-key-concepts/1-what-are-ai-agents.mdx b/docs/10-key-concepts/1-what-are-ai-agents.mdx
@@ -2,8 +2,4 @@
 
 ![](/img/screenshots/10-key-concepts/agent-workflow.png)
 
-An AI agent is a system that uses an LLM to **reason** through a problem, create a **plan** to solve the problem, and **execute** the plan with the help of a set of tools.
-
-In multi-agent systems, two or more agents collaborate or orchestrate and delegate tasks to solve problems.
-
-This way, agentic systems can handle complex, multi-step queries, and also self-revise and refine responses.  
+An AI agent is a system that uses an LLM to **reason** through a problem, create a **plan** to solve the problem, and **execute and iterate** on the plan with the help of a set of tools.
diff --git a/docs/10-key-concepts/4-agent-architectures.mdx b/docs/10-key-concepts/4-agent-architectures.mdx
@@ -0,0 +1,32 @@
+# 🦹 Agent architectures
+
+AI agents can consist of a single LLM as the decision maker or multiple LLMs (or same LLM, different prompts) working together to solve complex problems.
+
+Let's look at a few different agent architectures:
+
+## Tool-calling
+
+This is the architecture most agents start with. It consists of a single LLM that has access to several tools to perform a range of tasks.
+
+![](/img/screenshots/10-key-concepts/agent-workflow.png)
+
+If you start with tool-calling agents, but upon thorough evaluation find that you need a more sophisticated architecture, only then consider multi-agent architectures. Bear in mind that fully autonomous multi-agent workflows mean higher costs, latency, and a system that is hard to debug, so use them with caution. 
+
+## Supervisor
+
+In this architecture, a single agent (supervisor) interfaces with a group of agents to determine the next course of action.
+
+![](/img/screenshots/10-key-concepts/supervisor.png)
+
+## Network
+
+In this architecture, each agent can communicate with every other agent and decide which one to call next or end the execution.
+
+![](/img/screenshots/10-key-concepts/network.png)
+
+
+## Custom
+
+In this setup, you can decide which agents can interact with each other and how the control flows between them.
+
+![](/img/screenshots/10-key-concepts/custom.png)
diff --git a/static/img/screenshots/10-key-concepts/agent-workflow.png b/static/img/screenshots/10-key-concepts/agent-workflow.png
diff --git a/static/img/screenshots/10-key-concepts/custom.png b/static/img/screenshots/10-key-concepts/custom.png
diff --git a/static/img/screenshots/10-key-concepts/network.png b/static/img/screenshots/10-key-concepts/network.png
diff --git a/static/img/screenshots/10-key-concepts/supervisor.png b/static/img/screenshots/10-key-concepts/supervisor.png