Agentic AI 103: Constructing Multi-Agent Groups

Cell App Improvement with Python | In direction of Knowledge Science

Mastering SQL Window Capabilities | In the direction of Information Science

Introduction

articles right here in TDS, we explored the basics of Agentic AI. I’ve been sharing with you some ideas that may allow you to navigate by this sea of content material we have now been seeing on the market.

Within the final two articles, we explored issues like:

The best way to create your first agent
What are instruments and tips on how to implement them in your agent
Reminiscence and reasoning
Guardrails
Agent analysis and monitoring

Good! If you wish to examine it, I counsel you take a look at the articles [1] and [2] from the References part.

Agentic AI is among the hottest topics in the meanwhile, and there are a number of frameworks you possibly can select from. Thankfully, one factor that I’ve seen from my expertise studying about brokers is that these elementary ideas will be carried over from one to a different.

For instance, the category Agent from one framework turns into chat in one other, and even one thing else, however normally with comparable arguments and the exact same goal of connecting with a Massive Language Mannequin (LLM).

So let’s take one other step in our studying journey.

On this publish, we are going to learn to create multi-agent groups, opening alternatives for us to let AI carry out extra complicated duties for us.

For the sake of consistency, I’ll proceed to make use of Agno as our framework.

Let’s do that.

Multi-Agent Groups

A multi-agent group is nothing greater than what the phrase means: a group with multiple agent.

However why do we want that?

Nicely, I created this easy rule of thumb for myself that, if an agent wants to make use of greater than 2 or 3 instruments, it’s time to create a group. The explanation for that is that two specialists working collectively will do significantly better than a generalist.

While you attempt to create the “swiss-knife agent”, the likelihood of seeing issues going backwards is excessive. The agent will turn out to be too overwhelmed with completely different directions and the amount of instruments to take care of, so it finally ends up throwing an error or returning a poor consequence.

Then again, while you create brokers with a single goal, they may want only one device to resolve that downside, subsequently growing efficiency and enhancing the consequence.

To coordinate this group of specialists, we are going to use the category Staff from Agno, which is ready to assign duties to the right agent.

Let’s transfer on and perceive what we are going to construct subsequent.

Undertaking

Our challenge can be centered on the social media content material era trade. We are going to construct a group of brokers that generates an Instagram publish and suggests a picture primarily based on the subject offered by the person.

The person sends a immediate for a publish.
The coordinator sends the duty to the Author
- It goes to the web and searches for that matter.
The Author returns textual content for the social media publish.
As soon as the coordinator has the primary consequence, it routes that textual content to the Illustrator agent, so it will possibly create a immediate for a picture for the publish.

Workflow of the Staff of brokers. Picture by the creator.

Discover how we’re separating the duties very effectively, so every agent can focus solely on their job. The coordinator will be sure that every agent does their work, and they’re going to collaborate for a great remaining consequence.

To make our group much more performant, I’ll prohibit the topic for the posts to be created about Wine & Effective Meals. This manner, we slim down much more the scope of data wanted from our agent, and we are able to make its function clearer and extra centered.

Let’s code that now.

Code

First, set up the required libraries.

pip set up agno duckduckgo-search google-genai

Create a file for setting variables .env and add the wanted API Keys for Gemini and any search mechanism you’re utilizing, if wanted. DuckDuckGo doesn’t require one.

GEMINI_API_KEY="your api key"
SEARCH_TOOL_API_KEY="api key"

Import the libraries.

# Imports
import os
from textwrap import dedent
from agno.agent import Agent
from agno.fashions.google import Gemini
from agno.group import Staff
from agno.instruments.duckduckgo import DuckDuckGoTools
from agno.instruments.file import FileTools
from pathlib import Path

Creating the Brokers

Subsequent, we are going to create the primary agent. It’s a sommelier and specialist in connoisseur meals.

It wants a title for simpler identification by the group.
The function telling it what its specialty is.
A description to inform the agent tips on how to behave.
The instruments that it will possibly use to carry out the duty.
add_name_to_instructions is to ship together with the response the title of the agent who labored on that job.
We describe the expected_output.
The mannequin is the mind of the agent.
exponential_backoff and delay_between_retries are to keep away from too many requests to LLMs (error 429).

# Create particular person specialised brokers
author = Agent(
    title="Author",
    function=dedent("""
                You might be an skilled digital marketer who focuses on Instagram posts.
                You understand how to jot down an interesting, Web optimization-friendly publish.
                You recognize all about wine, cheese, and connoisseur meals present in grocery shops.
                You might be additionally a wine sommelier who is aware of tips on how to make suggestions.
                
                """),
    description=dedent("""
                Write clear, participating content material utilizing a impartial to enjoyable and conversational tone.
                Write an Instagram caption in regards to the requested {matter}.
                Write a brief name to motion on the finish of the message.
                Add 5 hashtags to the caption.
                When you encounter a personality encoding error, take away the character earlier than sending your response to the Coordinator.
                        
                        """),
    instruments=[DuckDuckGoTools()],
    add_name_to_instructions=True,
    expected_output=dedent("Caption for Instagram in regards to the {matter}."),
    mannequin=Gemini(id="gemini-2.0-flash-lite", api_key=os.environ.get("GEMINI_API_KEY")),
    exponential_backoff=True,
    delay_between_retries=2
)

Now, allow us to create the Illustrator agent. The arguments are the identical.

# Illustrator Agent
illustrator = Agent(
    title="Illustrator",
    function="You might be an illustrator who focuses on footage of wines, cheeses, and positive meals present in grocery shops.",
    description=dedent("""
                Based mostly on the caption created by Marketer, create a immediate to generate an interesting picture in regards to the requested {matter}.
                When you encounter a personality encoding error, take away the character earlier than sending your response to the Coordinator.
                
                """),
    expected_output= "Immediate to generate an image.",
    add_name_to_instructions=True,
    mannequin=Gemini(id="gemini-2.0-flash", api_key=os.environ.get("GEMINI_API_KEY")),
    exponential_backoff=True,
    delay_between_retries=2
)

Creating the Staff

To make these two specialised brokers work collectively, we have to use the category Agent. We give it a reputation and use the argument to find out the kind of interplay that the group may have. Agno makes obtainable the modes coordinate, route or collaborate.

Additionally, don’t neglect to make use of share_member_interactions=True to be sure that the responses will move easily among the many brokers. You may as well use enable_agentic_context, that permits group context to be shared with group members.

The argument monitoring is sweet if you wish to use Agno’s built-in monitor dashboard, obtainable at https://app.agno.com/

# Create a group with these brokers
writing_team = Staff(
    title="Instagram Staff",
    mode="coordinate",
    members=[writer, illustrator],
    directions=dedent("""
                        You're a group of content material writers working collectively to create participating Instagram posts.
                        First, you ask the 'Author' to create a caption for the requested {matter}.
                        Subsequent, you ask the 'Illustrator' to create a immediate to generate an interesting illustration for the requested {matter}.
                        Don't use emojis within the caption.
                        When you encounter a personality encoding error, take away the character earlier than saving the file.
                        Use the next template to generate the output:
                        - Submit
                        - Immediate to generate an illustration
                        
                        """),
    mannequin=Gemini(id="gemini-2.0-flash", api_key=os.environ.get("GEMINI_API_KEY")),
    instruments=[FileTools(base_dir=Path("./output"))],
    expected_output="A textual content named 'publish.txt' with the content material of the Instagram publish and the immediate to generate an image.",
    share_member_interactions=True,
    markdown=True,
    monitoring=True
)

Let’s run it.

# Immediate
immediate = "Write a publish about: Glowing Water and sugestion of meals to accompany."

# Run the group with a job
writing_team.print_response(immediate)

That is the response.

Picture of the Staff’s response. Picture by the creator.

That is how the textual content file seems like.

- Submit
Elevate your refreshment recreation with the effervescence of glowing water! 
Overlook the sugary sodas, and embrace the crisp, clear style of bubbles. 
Glowing water is the last word palate cleanser and a flexible companion for 
your culinary adventures.

Pair your favourite glowing water with connoisseur delights out of your native
grocery retailer.
Attempt these pleasant duos:

*   **For the Traditional:** Glowing water with a squeeze of lime, served with 
creamy brie and crusty bread.
*   **For the Adventurous:** Glowing water with a splash of cranberry, 
alongside a pointy cheddar and artisan crackers.
*   **For the Wine Lover:** Glowing water with a touch of elderflower, 
paired with prosciutto and melon.

Glowing water is not only a drink; it is an expertise. 
It is the right approach to get pleasure from these particular moments.

What are your favourite glowing water pairings?

#SparklingWater #FoodPairing #GourmetGrocery #CheeseAndWine #HealthyDrinks

- Immediate to generate a picture
A vibrant, eye-level shot inside a connoisseur grocery retailer, showcasing a variety
of glowing water bottles with varied flavors. Prepare pairings round 
the bottles, together with a wedge of creamy brie with crusty bread, sharp cheddar 
with artisan crackers, and prosciutto with melon. The lighting ought to be shiny 
and alluring, highlighting the textures and colours of the meals and drinks.

After we have now this textual content file, we are able to go to no matter LLM we like higher to create photos, and simply copy and paste the Immediate to generate a picture.

And here’s a mockup of how the publish can be.

Mockup of the publish generated by the Multi-agent group. Picture by the creator.

Fairly good, I’d say. What do you suppose?

Earlier than You Go

On this publish, we took one other step in studying about Agentic AI. This matter is sizzling, and there are numerous frameworks obtainable available in the market. I simply stopped making an attempt to be taught all of them and selected one to begin truly constructing one thing.

Right here, we have been capable of semi-automate the creation of social media posts. Now, all we have now to do is select a subject, alter the immediate, and run the Staff. After that, it’s all about going to social media and creating the publish.

Definitely, there may be extra automation that may be carried out on this move, however it’s out of scope right here.

Concerning constructing brokers, I like to recommend that you just take the simpler frameworks to begin, and as you want extra customization, you possibly can transfer on to LangGraph, for instance, which permits you that.