Agents Overview

Profound Agents combine AEO data, AI models, and connected tools into one platform, making it easy to automate analysis, generate insights, and take action.

Because Agents are highly versatile, their use cases are broad. Examples include:

• Running Answer Engine Insights reports and automatically sending them to your team via Slack.

• Identifying content gaps and automatically creating optimized content to fill those gaps.

• Detecting underperforming content and automatically updating it to perform better in AI search.

The Agents product isn’t designed to solve only a fixed set of use cases; instead, it’s built to adapt to your team’s unique needs.

You can use pre-built Agents, build custom Agents, or modify an existing Agent.

Components of a Agent

There are several key components common to all Agents. These elements define the Agent's structure, and a solid understanding of each will help you design custom Agents that meet your unique goals. 

• Inputs: Inputs are the information you provide each time you start an Agent. Inputs can include text, Profound data, files, or dropdown selections.

• Variables: Variables are values that can be passed between steps. The initial input becomes a variable, and each step’s output also becomes a variable that can be reused in subsequent steps.

• Nodes:  Nodes are the steps you add to an Agent. Each node performs a specific task and returns an output. Example nodes include Prompt LLM, Call API, and Answer Engine Insights.

• Agent Output: The Agent's final Output is the end result produced after all steps run.

Templates

Agent Templates are pre-built Agents that address common scenarios. You can use a template as-is or treat it as a starting point for a custom Agent. 

Popular templates include:

• AEO Content Refresh

• Generate AEO-Optimized FAQs

• Query Fanout Estimator

• Generate Blog Post from YouTube Video

Starter Template

The Starter Template is designed to help you get familiar with Agents and their core components. At a high level, it takes an input and uses it as a ChatGPT prompt.

The Starter Template includes the following steps: Input → Prompt LLM → Output.

• The Input block captures a term that you specify each time the Agent runs.

• The Agent passes this input to the Prompt LLM node, which prompts Chat-GPT to summarize the text.

• The engine’s response is then delivered as the Output.

While the template contains only a few steps, it provides a solid foundation to build on. For example, you can take the ChatGPT output and use it in a custom API call or generate new custom content using the returned information.

Agent Runs

An Agent run is a single execution of an Agent. While the Agent itself acts as a template, each time it processes an input, it creates a distinct Agent run.