	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Agentic AI > CrewAI > CrewAI Multi-Agent Research

CrewAI Multi-Agent Research

Author: Venkata Sudhakar

CrewAI lets you build teams of specialised AI agents where each agent has a defined role, goal, and backstory shaping how it approaches work. Instead of one agent doing everything, you compose a crew: a researcher who gathers information, an analyst who interprets it, and a writer who produces the polished output. Each agent works on its task sequentially, passing output to the next. The crew pattern mirrors how human teams actually work - specialists collaborating with clear handoffs rather than a single generalist trying to do everything at once.

A CrewAI crew has three building blocks: agents (specialists with roles, goals, backstories, and optional tools), tasks (specific work items with expected outputs assigned to agents), and a crew object that orchestrates the execution. Tasks set to sequential process run one after another, with each task automatically receiving the previous task output as context. Agents can use tools like web search, file reading, or code execution autonomously as needed to complete their tasks.

The below example builds a three-agent competitive analysis crew for a SaaS product team: a researcher gathering competitor data, a strategy analyst identifying opportunities and threats, and a report writer producing an executive brief the CEO can act on.

from crewai import Agent, Task, Crew, Process
from crewai_tools import SerperDevTool
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-4o-mini", api_key="your-api-key", temperature=0.3)
search_tool = SerperDevTool(api_key="your-serper-api-key")

# Agent 1: Market Researcher
researcher = Agent(
    role="Senior Market Research Analyst",
    goal=("Gather factual information about competitor products, "
          "pricing, features and recent announcements for Indian SaaS."),
    backstory=(
        "Expert market researcher with 10 years in Indian B2B SaaS. "
        "You find accurate, up-to-date information from public sources "
        "and present it factually without bias."
    ),
    tools=[search_tool],
    llm=llm, verbose=True
)

# Agent 2: Strategy Analyst
analyst = Agent(
    role="Business Strategy Analyst",
    goal=("Analyse competitor research to identify strategic opportunities, "
          "threats, and positioning gaps for the client company."),
    backstory=(
        "Sharp strategy consultant who turns raw market data into actionable "
        "insights. You think in competitive advantages and market gaps."
    ),
    llm=llm, verbose=True
)

# Agent 3: Executive Report Writer
writer = Agent(
    role="Executive Report Writer",
    goal="Write a clear, concise competitive analysis report a CEO can act on.",
    backstory=(
        "You write executive-level reports that are data-driven and concise. "
        "You end every report with prioritised actionable recommendations. "
        "You never use vague filler statements."
    ),
    llm=llm, verbose=True
)

Defining tasks and running the crew on a real business analysis request,

COMPANY = "Zoho CRM"
COMPETITORS = "Salesforce, HubSpot, Freshsales"

research_task = Task(
    description=(
        "Research the top competitors of " + COMPANY + " in the Indian CRM market. "
        "Analyse: " + COMPETITORS + ". "
        "For each: current pricing in INR, key features, 2024-2025 product updates, "
        "customer review sentiment, and India-specific presence. Facts only."
    ),
    expected_output=(
        "Structured research with one section per competitor covering "
        "pricing, features, recent updates, and India presence."
    ),
    agent=researcher
)

analysis_task = Task(
    description=(
        "Using the research, analyse the competitive position of " + COMPANY + ". "
        "Identify: where " + COMPANY + " has clear advantages, where competitors "
        "are stronger, underserved segments, top 3 threats and top 3 opportunities."
    ),
    expected_output=(
        "Strategic analysis with sections for advantages, weaknesses, "
        "gaps, threats, and opportunities - each backed by specific evidence."
    ),
    agent=analyst,
    context=[research_task]
)

report_task = Task(
    description=(
        "Write an executive competitive analysis report for " + COMPANY + " leadership. "
        "Structure: Executive Summary (3 bullets), Competitor Snapshots (1 para each), "
        "Strategic Position, Top 5 Prioritised Actions. Under 600 words."
    ),
    expected_output=(
        "Polished executive report under 600 words with specific numbers "
        "and 5 prioritised action items."
    ),
    agent=writer,
    context=[research_task, analysis_task]
)

crew = Crew(
    agents=[researcher, analyst, writer],
    tasks=[research_task, analysis_task, report_task],
    process=Process.sequential,
    verbose=True
)

result = crew.kickoff()
print("=== FINAL COMPETITIVE ANALYSIS REPORT ===")
print(result)

It gives the following condensed output showing agent handoffs,

[Researcher] Searching: Zoho CRM pricing India 2025...
[Researcher] Searching: Salesforce India CRM pricing features 2025...
[Researcher] Searching: HubSpot CRM India pricing 2025...
[Researcher] Research complete. Handing off to Analyst.

[Analyst] Analysing competitive gaps and strategic position...
[Analyst] Identified 3 key opportunities and 3 threats. Handing off to Writer.

[Writer] Drafting executive report...

=== FINAL COMPETITIVE ANALYSIS REPORT ===
EXECUTIVE SUMMARY
- Zoho CRM leads on price-value ratio in India at Rs 1,300/user/month vs
  Salesforce at Rs 5,800+ - a 4x cost advantage for SMB customers.
- Salesforce and HubSpot are expanding India enterprise presence with
  dedicated data centres and local support teams in 2025.
- Key opportunity: mid-market segment (50-500 employees) is underserved
  by all three competitors on vernacular language support.

COMPETITOR SNAPSHOTS
Salesforce: Dominant in enterprise. Recent 2025 launch of Agentforce AI
  platform threatens to widen feature gap at the top end.
HubSpot: Strong inbound marketing + CRM integration. Free tier drives
  top-of-funnel adoption that converts to paid aggressively.
Freshsales: Closest direct competitor. Fresher UI, aggressive India pricing,
  but weaker analytics than Zoho.

TOP 5 PRIORITISED ACTIONS
1. Launch Hindi/regional language interface - capture underserved SMB segment
2. Build AI assistant feature parity with Salesforce Agentforce
3. Create HubSpot migration tool to capture their churning free-tier users
4. Publish India-specific ROI calculator to counter enterprise deals
5. Expand Chennai/Pune support centres for regional enterprise credibility

# Researcher, Analyst, Writer each did their job independently
# Full report produced autonomously from a single crew.kickoff() call

CrewAI works best when tasks genuinely require different expertise or perspectives. A research task needs factual grounding and tool use; an analysis task needs strategic interpretation; a writing task needs clarity and structure. If all three can be done well by the same system prompt, use a single agent instead. For longer running crews, set memory=True on the crew to allow agents to remember relevant context across tasks. CrewAI also supports hierarchical process where a manager agent can delegate tasks dynamically - useful when you do not know in advance exactly how many research subtasks will be needed.

Send your comments, suggestions or queries regarding this site to [email protected].