Skip to main content

Agents

Some applications will require not just a predetermined chain of calls to LLMs/other tools, but potentially an unknown chain that depends on the user's input. In these types of chains, there is a β€œagent” which has access to a suite of tools. Depending on the user input, the agent can then decide which, if any, of these tools to call.

At the moment, there are two main types of agents:

  1. Action Agents: these agents decide an action to take and take that action one step at a time
  2. Plan-and-Execute Agents: these agents first decide a plan of actions to take, and then execute those actions one at a time.

When should you use each one?

  • Action Agents are more conventional, and good for small tasks.
  • For more complex or long running tasks, the initial planning step of Plan-and-Execute Agents helps to maintain long term objectives and focus, at the expense of generally more calls and higher latency.

These two agents are also not mutually exclusive - in fact, it is often best to have an Action Agent be in charge of the execution for the Plan and Execute agent.

Action Agents​

The high-level pseudocode of an Action Agent looks something like:

  • Some user input is received
  • The agent decides which tool - if any - to use, and what the input to that tool should be
  • That tool is then called with that tool input, and an observation is recorded (this is just the output of calling that tool with that tool input).
  • That history of tool, tool input, and observation is passed back into the agent, and it decides what steps to take next
  • This is repeated until the agent decides it no longer needs to use a tool, and then it responds directly to the user.
interface AgentStep {
action: AgentAction;
observation: string;
}

interface AgentAction {
tool: string; // Tool.name
toolInput: string; // Tool.call argument
}

interface AgentFinish {
returnValues: object;
}

class Agent {
plan(steps: AgentStep[], inputs: object): Promise<AgentAction | AgentFinish>;
}

Plan-and-Execute Agents​

The high level pseudocode of a Plan-and-Execute Agent looks something like:

  • Some user input is received
  • The planner lists out the steps to take
  • The executor goes through the list of steps, executing them one-by-one until outputting the final result

The current implementation is to use an LLMChain as the planner and an Action Agent as the executor.

Go deeper​