Example - CLI
Example: AI-Driven Market Research (Using the the CLI)
To better understand how to run a workflow using Peeps AI, let's walk through a practical example. This example demonstrates how to set up a simple multi-agent system to automate a market research task.
Project Setup
Start by creating a new Peeps AI project:
peepsai create group market_research
This command generates the basic folder structure with necessary configuration files.
Defining Agents and Tasks
agents.yaml
Define two agents: a Researcher and a Data Analyst.
# src/market_research/config/agents.yaml
researcher:
role: "Market Research Specialist"
goal: "Identify key trends in the AI industry."
backstory: "An expert with years of experience analyzing industry trends and market data."
data_analyst:
role: "Data Analyst"
goal: "Analyze collected data and generate actionable insights."
backstory: "Proficient in data interpretation and presenting findings clearly."
tasks.yaml
Outline the tasks for each agent.
# src/market_research/config/tasks.yaml
collect_data:
description: "Research the latest AI trends for 2024."
expected_output: "A list of key trends and emerging technologies."
agent: researcher
analyze_data:
description: "Analyze the research findings to identify actionable insights."
expected_output: "A detailed report summarizing key trends with recommendations."
agent: data_analyst
output_file: insights_report.md
Customizing the Group Logic
Modify group.py to define the workflow process:
# src/market_research/group.py
from peepsai import Agent, Peeps, Process, Task
from peepsai.project import PeepsBase, agent, group, task
@PeepsBase
class MarketResearchGroup:
@agent
def researcher(self) -> Agent:
return Agent(config=self.agents_config['researcher'])
@agent
def data_analyst(self) -> Agent:
return Agent(config=self.agents_config['data_analyst'])
@task
def collect_data(self) -> Task:
return Task(config=self.tasks_config['collect_data'])
@task
def analyze_data(self) -> Task:
return Task(config=self.tasks_config['analyze_data'])
@group
def group(self) -> Peeps:
return Peeps(
agents=self.agents,
tasks=self.tasks,
process=Process.sequential
)
Running the Workflow
Navigate to the project directory:
cd market_research
Execute the workflow:
peepsai run
Alternatively, run it directly with Python:
python src/market_research/main.py
Reviewing the Output
After successful execution:
Check the console output for real-time logs of each task's progress.
Locate the final report in
insights_report.md
generated by the Data Analyst.
Advanced Example - Adding Dynamic Inputs
You can pass custom inputs to make the workflow dynamic:
peepsai run --inputs '{"industry": "Healthcare AI"}'
Update agents.yaml and tasks.yaml to utilize the {industry}
variable dynamically in descriptions and goals.
Troubleshooting Tips
If the workflow doesn't start:
peepsai run --verbose
This will display detailed logs for debugging.
Missing dependencies:
pip install 'peepsai[tools]'
C
Last updated