I Built an Entire Marketing Team with 37 AI Agents (Free Template)

Key Concepts

AI Agents: Autonomous entities designed to perform specific tasks.
Multi-Agent Teams: Groups of AI agents working together to achieve a common goal.
Deterministic vs. Non-Deterministic Automation: Deterministic automation follows pre-defined rules, while non-deterministic automation uses AI to make decisions.
Tools: Deterministic workflows that provide AI agents with access to software and specific functionalities.
Workforces: Combinations of AI agents and tools organized to execute complex tasks.
Relevance AI Chat Feature: A new feature allowing users to interact with individual AI agents within a unified chat interface or Slack.
Relevance AI Marketplace: A platform for sharing and selling AI agents and tools.
Human-in-the-Loop: Incorporating human oversight and intervention in AI-driven processes.
Deep Research: Leveraging AI to conduct in-depth research on specific topics.
Topic Clustering: Creating a network of related articles around a central theme for SEO purposes.

37 AI Agent Marketing and Sales Team Overview

The video demonstrates a 37 AI agent marketing and sales team built within Relevance AI. These agents are designed to automate various tasks, including content creation, SEO, lead qualification, and sales outreach. The presenter emphasizes the new chat feature in Relevance AI, which allows direct interaction with individual agents, addressing the "broken telephone" problem often encountered in multi-agent systems.

Demo of the Chat Feature

The presenter showcases the chat feature by repurposing a YouTube video. He tags the LinkedIn post writer, newsletter writer, and X (formerly Twitter) writer agents, providing the video link and instructing them to create posts. The agents generate content tailored to each platform's style.

Example: The LinkedIn post writer creates a post with a call to action, while the newsletter agent generates a longer, more personal newsletter. The X writer creates a shorter, punchier post.
Image Generation: The presenter then tags the image generator agent to create an image of himself in Amsterdam and an infographic for the LinkedIn post. The infographic highlights the "three trillion SAS agencies" opportunity.
Agent Memory: The presenter demonstrates the ability to add memory to agents by instructing the X writer agent to remember a specific post as a reference for future content generation.

Deep Research and Presentation Generation

The presenter demonstrates using OpenAI's deep research feature in conjunction with agents. He instructs the deep researcher to conduct research on optimizing websites for AI search. The results are then used by the presentation creator agent to generate a slideshow for marketing agencies.

Process: The deep researcher takes approximately five minutes to gather insights. The presentation agent then creates a presentation with an agenda covering AI search fundamentals, ranking signals, and content/technical optimization.
Tool Integration: The presentation is generated using GMA (AI presentation generator), showcasing the integration of external tools.

SEO Workforce Demo

The presenter demonstrates the SEO workforce, a team of agents designed to create multiple articles in a topic cluster. This involves switching Relevance AI to "workforce mode" and interacting with the manager agent.

Process: The manager agent delegates tasks to the keyword researcher (using Moz API for keyword analytics), content outliner, content writer, and blog asset maker. The team generates multiple articles around the topic of optimizing for AI search.
Example: The SEO team generates articles on "using AI for content optimization," "how to use AI tools for better SEO," and "the future of AI SEO optimization trends and predictions."
Timeframe: The entire process takes approximately 35 minutes.

Slack Integration Demo

The presenter demonstrates the ability to access AI agents within Slack. He tags Relevance AI and then a specific agent (Quinn, the lead qualifier agent) to research a lead.

Process: The lead qualifier agent researches the lead (Sarah from Slack.com) and updates HubSpot with the contact information and a follow-up task.
Benefits: The Slack integration promotes agent adoption within the organization, as team members can easily access and observe the use of AI agents.

Overview of Relevance AI

The presenter contrasts Relevance AI with Naden, highlighting that Relevance AI focuses on non-deterministic automation (AI agents), while Naden focuses on deterministic workflows. He notes that deterministic systems are currently more reliable for client-facing applications but acknowledges the potential of Relevance AI's new features for marketing and sales use cases.

Relevance AI New Features and Functionality

Marketplace: A platform for sharing and selling AI agents and tools.
Tasks: A feature for tracking tasks across the organization.
Tools: Deterministic workflows that provide AI agents with access to software.
- Tool Building: Tools can be built manually or automatically using the "invent" feature.
- Example: The presenter demonstrates building a tool that generates an image with nano banana using the "invent" feature.
Agents: Autonomous entities with prompts, tools, knowledge bases, and triggers.
- Agent Creation: Agents can be created manually or automatically using the "magic instruction" feature.
- Memory: Agents can be given memory to remember specific information or preferences.
- Escalations: Human approval can be required before an agent takes certain actions.
Workforces: Combinations of AI agents and tools organized to execute complex tasks.
- Connections: Agents and tools can be connected using AI connections (sub-agents) or next-step handoffs (workflows).
- Triggers: Workforces can be triggered manually or externally (e.g., from Slack).
- Conditions: Conditions can be set to determine when certain steps are executed.
Integrations: Relevance AI now offers thousands of integrations with various software platforms.

Importing and Setting Up the Agents

The presenter provides a link to a free resource in his community where users can access a link to import the 37 AI agents into their Relevance AI account. He emphasizes that the agents are designed to be used individually or in small groups via the chat feature, rather than as a single, complex multi-agent system. Users will need to update the agents with their own API keys and software connections.

Slack Integration Setup

The presenter demonstrates how to connect Relevance AI to Slack. This involves connecting the Relevance AI account to Slack and adding the Relevance AI app to specific Slack channels. Once connected, users can trigger AI agents by tagging Relevance AI followed by a keyword.

Conclusion

The video provides a comprehensive overview of a 37 AI agent marketing and sales team built within Relevance AI. The presenter highlights the new chat feature, which allows direct interaction with individual agents, and demonstrates various use cases, including content creation, SEO, lead qualification, and sales outreach. He also provides a detailed explanation of Relevance AI's features and functionality, including tools, agents, workforces, and integrations. The video concludes with instructions on how to import the agents and set up the Slack integration. The main takeaway is that Relevance AI, with its new features, offers a compelling platform for building and deploying AI agents for various business tasks, particularly in marketing and sales, and that the human-in-the-loop approach facilitated by the chat and Slack integrations can help to mitigate the risks associated with fully autonomous AI systems.