Added gradio UI

.env.example

@@ -1,2 +1,6 @@
 MODEL=gemini/gemini-1.5-flash
-GEMINI_API_KEY=<gemini_api_key> # Your API key here
+GEMINI_API_KEY=<gemini_api_key> # Your API key here
+NEO_DB_URI = <neo4j_uri> # Your URI here
+NEO_DB_USERNAME = <neo4j_username> # Your username here
+NEO_DB_PWD = <neo4j_password> # Your password here
+NEO_DB_DATABSE = <neo4j_database> # Your database here

knowledge/cities.json

@@ -0,0 +1,104 @@
+{
+  "cities": [
+    "Seattle",
+    "Rome",
+    "Manchester",
+    "Schimmert",
+    "Buenos Aires",
+    "Chicago",
+    "New York City",
+    "Redwood City",
+    "Sydney",
+    "Auckland",
+    "Brisbane",
+    "Neuilly",
+    "Munich",
+    "Milan",
+    "Paris",
+    "Shanghai",
+    "Longmont",
+    "Houston",
+    "Tampa",
+    "Chattanooga",
+    "Philadelphia",
+    "Visalia",
+    "Dinard",
+    "London",
+    "Fremont",
+    "Pleasanton",
+    "Bangalore",
+    "Toronto",
+    "Lausanne",
+    "Paceco",
+    "Horgen",
+    "Makhachkala",
+    "Mozzate",
+    "Jaux",
+    "Porto",
+    "Kuala Lumpur",
+    "Torredembarra",
+    "Madrid",
+    "Columbia",
+    "San Francisco",
+    "Copenhagen",
+    "Perth",
+    "West Perth",
+    "Weehawken",
+    "Montreal",
+    "Quebec City",
+    "Saint-Priest-de-Gimel",
+    "Saint-Priest",
+    "Singapore",
+    "Jersey City",
+    "Würzburg",
+    "Düsseldorf",
+    "Urbandale",
+    "Naples",
+    "Hilversum",
+    "São Paulo",
+    "St. Albert",
+    "San Diego",
+    "Ronkonkoma",
+    "Chatswood",
+    "Conshohocken",
+    "Overland Park",
+    "Victoria",
+    "Meridian",
+    "Irvine",
+    "Atlanta",
+    "Dallas",
+    "Paradise Valley",
+    "Prescott Valley",
+    "Plano",
+    "Wayne",
+    "Ottawa",
+    "Gilbert",
+    "Milwaukee",
+    "Los Angeles",
+    "Boston",
+    "Wellington",
+    "Eden Prairie",
+    "Groningen",
+    "Amsterdam",
+    "Kyiv",
+    "Dnipro",
+    "Ahmedabad",
+    "Hollywood",
+    "Portland",
+    "Carlsbad",
+    "Columbus",
+    "Bentonville",
+    "Charlotte",
+    "The Rocks",
+    "Chesterfield",
+    "Rio de Janeiro",
+    "Créteil",
+    "Oakland",
+    "Phoenix",
+    "Indianapolis",
+    "Best",
+    "Raleigh",
+    "Pune",
+    "Durham"
+  ]
+}

knowledge/content_style_mapping.json

@@ -1,48 +0,0 @@
-{
-    "target_audience": {
-      "linkedin_post": {
-        "format": "post",
-        "tone": "professional"
-      },
-      "whatsapp_message": {
-        "format": "chat",
-        "tone": "casual"
-      },
-      "tweet": {
-        "format": "tweet",
-        "tone": "straightforward"
-      },
-      "news_article": {
-        "format": "article",
-        "tone": "neutral"
-      },
-      "technical_blog": {
-        "format": "blog",
-        "tone": "professional"
-      },
-      "formal_email": {
-        "format": "email",
-        "tone": "professional"
-      },
-      "instagram_post": {
-        "format": "post",
-        "tone": "friendly"
-      },
-      "website_content": {
-        "format": "report",
-        "tone": "neutral"
-      },
-      "marketing_email": {
-        "format": "email",
-        "tone": "confident"
-      },
-      "job_application": {
-        "format": "email",
-        "tone": "professional"
-      },
-      "customer_support_response": {
-        "format": "chat",
-        "tone": "friendly"
-      }
-    }
-  }

knowledge/format.json

@@ -1,46 +0,0 @@
-{
-  "format": {
-    "post": {
-      "name": "Post",
-      "description": "A short and engaging content piece, often used for platforms like LinkedIn or Facebook. It highlights key points and encourages interaction, such as likes, shares, or comments.",
-      "max_length": 300,
-      "negative": "Avoid long paragraphs, excessive technical jargon, or overly formal language. Do not make the content too detailed or complex."
-    },
-    "chat": {
-      "name": "Chat",
-      "description": "A conversational and interactive response that mimics real-time messaging. It focuses on direct, clear, and natural communication with a friendly tone.",
-      "max_length": 200,
-      "negative": "Do not use overly formal language, complex sentence structures, or impersonal tones. Avoid too much technical information and hashtags."
-    },
-    "tweet": {
-      "name": "Tweet",
-      "description": "A concise and impactful message designed for Twitter or similar platforms. It often uses hashtags, mentions, or trending topics to increase visibility.",
-      "max_length": 280,
-      "negative": "Avoid lengthy explanations, irrelevant details, or complex sentences. Do not exceed the character limit or use a formal tone."
-    },
-    "email": {
-      "name": "Email",
-      "description": "A formal or semi-formal written message for direct communication. It includes a clear subject line, opening, body, and closing, often addressing specific recipients.",
-      "max_length": 1500,
-      "negative": "Avoid casual or overly brief content. Do not make the email too long or include excessive details that deviate from the purpose."
-    },
-    "blog": {
-      "name": "Blog",
-      "description": "A detailed and informative piece written for online readers. It includes a compelling introduction, multiple sections with headings, and a conclusion to engage and educate the audience.",
-      "max_length": 2000,
-      "negative": "Avoid overly technical language, jargon, or too formal language. Do not make the blog too short or lacking in detail."
-    },
-    "article": {
-      "name": "Article",
-      "description": "A comprehensive, well-researched write-up, often featuring in-depth analysis or expert perspectives. Suitable for magazines, websites, or journals to inform or persuade a broad audience.",
-      "max_length": 3000,
-      "negative": "Do not oversimplify complex topics, avoid using unverified sources or vague statements. Avoid making the article too conversational or informal."
-    },
-    "report": {
-      "name": "Report",
-      "description": "A structured document that presents facts, data, and analysis for a specific audience or purpose. It often includes charts, tables, and a clear summary of findings.",
-      "max_length": 5000,
-      "negative": "Avoid informal language, speculative content, or a lack of data. Do not make the report subjective or unorganized."
-    }
-  }
-}

knowledge/industries.json

@@ -0,0 +1,54 @@
+{
+  "industries": [
+    "Technology",
+    "Healthcare",
+    "Finance",
+    "Education",
+    "Retail",
+    "Manufacturing",
+    "Transportation",
+    "Entertainment",
+    "Real Estate",
+    "Energy",
+    "Telecommunications",
+    "Hospitality",
+    "Food & Beverage",
+    "Automotive",
+    "Aerospace",
+    "Biotechnology",
+    "Pharmaceuticals",
+    "Construction",
+    "Consulting",
+    "Media",
+    "Agriculture",
+    "Legal",
+    "Insurance",
+    "E-commerce",
+    "Cybersecurity",
+    "Sports",
+    "Logistics",
+    "Government",
+    "Nonprofit",
+    "Mining",
+    "Waste Management",
+    "Textile",
+    "Shipping",
+    "Marketing",
+    "Tourism",
+    "Advertising",
+    "Electronics",
+    "Defense",
+    "Architecture",
+    "Music",
+    "Publishing",
+    "Gaming",
+    "Social Media",
+    "Investment",
+    "Artificial Intelligence",
+    "Blockchain",
+    "Cloud Computing",
+    "Robotics",
+    "Nanotechnology",
+    "Environmental Services"
+  ]
+}

knowledge/target_audience.json

@@ -1,59 +0,0 @@
-{
-    "target_audience": {
-      "linkedin_post": {
-        "name": "LinkedIn Post",
-        "description": "Aimed at professionals, thought leaders, and businesses. Content is typically career-focused or industry-related, designed to spark engagement among people seeking professional growth, networking, or knowledge.",
-        "ideal_audience": "Business professionals, marketers, entrepreneurs, recruiters, job seekers, industry experts"
-      },
-      "whatsapp_message": {
-        "name": "WhatsApp Message",
-        "description": "Casual or semi-formal communication meant for direct, personal messaging. Used for sending quick updates, reminders, or messages to small groups, often informal in tone.",
-        "ideal_audience": "Friends, family, small teams, personal contacts, colleagues in a more relaxed setting"
-      },
-      "tweet": {
-        "name": "Tweet",
-        "description": "Aimed at a broad, diverse audience on Twitter, from the general public to niche communities. Tweets are brief and designed to provoke quick engagement, whether through likes, retweets, or comments.",
-        "ideal_audience": "General public, influencers, content creators, tech enthusiasts, trend followers, entertainment fans"
-      },
-      "news_article": {
-        "name": "News Article",
-        "description": "Targeted at a wide audience seeking timely, relevant information. News articles are often written for readers who are looking for the latest updates, in-depth analysis, or expert opinions on current events or trending topics.",
-        "ideal_audience": "General public, journalists, news enthusiasts, academics, policy makers, professionals interested in current affairs"
-      },
-      "technical_blog": {
-        "name": "Technical Blog",
-        "description": "Aimed at professionals, experts, or hobbyists in specialized fields like technology, software development, engineering, or data science. The content is typically in-depth, offering insights, tutorials, and technical knowledge.",
-        "ideal_audience": "Software developers, engineers, data scientists, tech entrepreneurs, IT professionals, learners seeking advanced technical skills"
-      },
-      "formal_email": {
-        "name": "Formal Email",
-        "description": "Designed for business or professional communication. The email is meant for communicating with colleagues, clients, business partners, or supervisors in a polite, respectful, and formal tone.",
-        "ideal_audience": "Business professionals, corporate partners, clients, managers, senior leadership, job candidates"
-      },
-      "instagram_post": {
-        "name": "Instagram Post",
-        "description": "Aimed at a visually-driven, creative audience. Instagram posts are often used for branding, lifestyle content, promotions, or personal expression, with a focus on imagery and concise captions.",
-        "ideal_audience": "Millennials, Gen Z, influencers, lifestyle bloggers, fashion enthusiasts, foodies, beauty influencers, brand followers"
-      },
-      "website_content": {
-        "name": "Website Content",
-        "description": "Targeted at a broad range of users, from potential customers to casual visitors. Website content needs to be informative, engaging, and easily navigable, with a focus on conversion and user experience.",
-        "ideal_audience": "Consumers, visitors, potential customers, search engine users, anyone looking for information or services"
-      },
-      "marketing_email": {
-        "name": "Marketing Email",
-        "description": "Aimed at potential customers or existing leads. Marketing emails are designed to drive action, whether it’s to purchase a product, sign up for a service, or engage with a special offer.",
-        "ideal_audience": "Prospective customers, leads, subscribers, clients, consumers interested in sales promotions"
-      },
-      "job_application": {
-        "name": "Job Application",
-        "description": "Directed towards hiring managers, recruiters, or HR professionals. The job application content should clearly convey qualifications, skills, and the applicant's interest in a specific role.",
-        "ideal_audience": "Recruiters, hiring managers, HR professionals, employers, staffing agencies"
-      },
-      "customer_support_response": {
-        "name": "Customer Support Response",
-        "description": "Meant for existing or potential customers seeking assistance with a product or service. The tone should be empathetic, understanding, and solution-oriented, addressing the customer’s issue in a clear, helpful manner.",
-        "ideal_audience": "Customers with inquiries, complaints, technical issues, or service requests"
-      }
-    }
-  }

knowledge/tone.json

@@ -1,36 +0,0 @@
-{
-    "tone": {
-      "professional": {
-        "name": "Professional",
-        "description": "Formal and respectful, suited for workplace communication."
-      },
-      "casual": {
-        "name": "Casual",
-        "description": "Relaxed and conversational, suitable for informal interactions."
-      },
-      "straightforward": {
-        "name": "Straightforward",
-        "description": "Direct and concise, avoiding unnecessary details."
-      },
-      "confident": {
-        "name": "Confident",
-        "description": "Assertive and positive, showing conviction in the message."
-      },
-      "friendly": {
-        "name": "Friendly",
-        "description": "Warm and approachable, making the user feel comfortable."
-      },
-      "neutral": {
-        "name": "Neutral",
-        "description": "Objective and balanced, without bias or emotion."
-      },
-      "storytelling": {
-        "name": "Storytelling",
-        "description": "Narrative style, weaving details into a compelling story."
-      },
-      "inspirational": {
-        "name": "Inspirational",
-        "description": "Uplifting and encouraging, motivating the user."
-      }
-    }
-  }

pyproject.toml

@@ -6,12 +6,14 @@ authors = [{ name = "Deepak Pant", email = "deepak.93p@gmail.com" }]
 requires-python = ">=3.10,<=3.13"
 dependencies = [
     "crewai[tools]>=0.86.0,<1.0.0",
-    "gradio>=5.12.0,<5.13.0"
+    "gradio>=5.12.0,<5.13.0",
+    "neo4j>=5.28.1,<6.0.0",
 ]
 
 [project.scripts]
 report_genie = "report_genie.app:launch"
 run_crew = "report_genie.main:run"
+# run_crew = "report_genie.app:launch"
 train = "report_genie.main:train"
 replay = "report_genie.main:replay"
 test = "report_genie.main:test"

src/report_genie/app.py

@@ -1,131 +1,57 @@
 import gradio as gr
 import os
-from report_genie.crew import ExpresslyServer
+from report_genie.crew import ReportGenieServer
+from report_genie.utils.utils import load_json_data
 
 
-def call(prompt, target_audience, format, tone, active_tab):
-    """
-    Calls the Expressly Server API to generate content based on the given inputs.
-
-    Args:
-    prompt (str): The text prompt for the chat.
-    target_audience (str): The target audience for the response.
-    format (str): The format of the response, e.g., text, markdown.
-    tone (str): The tone of the response, e.g., formal, informal.
-    active_tab (str): The active tab on the UI, either "target_audience" or "format_tone".
-
-    Returns:
-    str: The generated text response.
-
-    Raises:
-    ValueError: If prompt is empty, or if the active_tab value is invalid.
-    """
-
-    # Validating and constructing the inputs
-    if prompt is None or prompt == "":
-        raise ValueError("Prompt is required")
+# Load the JSON files
+CITIES_JSON_FILE = "cities.json"
+INDUSTRIES_JSON_FILE = "industries.json"
+KNOWLEDGE_SOURCE_PATH = "knowledge"
 
-    if active_tab == "target_audience":
-        format = ""
-        tone = ""
-    elif active_tab == "format_tone":
-        target_audience = ""
-    else:
-        raise ValueError("Invalid active_tab value")
+cities = load_json_data(CITIES_JSON_FILE, KNOWLEDGE_SOURCE_PATH)
+industries = load_json_data(INDUSTRIES_JSON_FILE, KNOWLEDGE_SOURCE_PATH)
 
+def generate_report(city, industry):
+ 
     inputs = {
-        "prompt": prompt,
-        "target_audience": target_audience,
-        "format": format,
-        "tone": tone,
+        "city_name": city,
+        "industry_name": industry,
     }
 
-    outputs = ExpresslyServer().crew().kickoff(inputs=inputs)
+    try:
+        output = ReportGenieServer().crew().kickoff(inputs=inputs)
+    except Exception:
+        output = None
 
-    if outputs is None:
+    if output is None:
         result = "Please check the inputs and try again. If the issue persists, contact support."
     else:
-        result = outputs.raw
+        result = output.raw
 
     return result
 
 
+## Gradio UI
 with gr.Blocks() as app:
-    gr.Markdown("# Expressly - Text Transformation App")
-
+    gr.Markdown("# 📄 Report Genie - An AI-powered automatic report generator")
+    gr.Markdown("Report Genie is an AI-driven report generation tool that automates the process of creating detailed and structured reports. Leveraging the power of **CrewAI** for task delegation and **Gradio** for an interactive user interface, this application streamlines report generation with minimal user input.")
+    
     with gr.Row():
-        # Left Column for Inputs
-        with gr.Column(scale=1):
-            prompt = gr.Textbox(label="Message Expressly", max_length=1024, lines=3)
-
-            # Create a state variable to store active tab
-            active_tab = gr.State("target_audience")
-
-            with gr.Tab("Target Audience", id="tab_audience") as tab1:
-                target_audience = gr.Dropdown(
-                    [
-                        "LinkedIn Post",
-                        "WhatsApp Message",
-                        "Tweet",
-                        "News Article",
-                        "Technical Blog",
-                        "Formal Email",
-                        "Instagram Post",
-                        "Website Content",
-                        "Marketing Email",
-                        "Job Application",
-                        "Customer Support Response",
-                    ],
-                    label="Target Audience",
-                    info="Pick a Target Audience to specify the purpose or platform.",
-                )
-                # Update state when this tab is selected
-                tab1.select(lambda: "target_audience", None, active_tab)
-
-            with gr.Tab("Format & Tone", id="tab_format") as tab2:
-                format = gr.Dropdown(
-                    [
-                        "Post",
-                        "Chat",
-                        "Tweet",
-                        "Email",
-                        "Blog",
-                        "Article",
-                        "Report",
-                        "Product Description",
-                    ],
-                    label="Format",
-                    info="Choose a Format to define the type of content.",
-                )
-                tone = gr.Dropdown(
-                    [
-                        "Professional",
-                        "Casual",
-                        "Straightforward",
-                        "Confident",
-                        "Friendly",
-                        "Neutral",
-                        "Storytelling",
-                        "Inspirational",
-                    ],
-                    label="Tone",
-                    info="Select a Tone to set the communication style.",
-                )
-                # Update state when this tab is selected
-                tab2.select(lambda: "format_tone", None, active_tab)
-
-            btn_submit = gr.Button("Submit")
-
-        # Right Column for Output
-        with gr.Column(scale=1):
-            results = gr.Markdown(label="Result")
-
-    btn_submit.click(
-        fn=call,
-        inputs=[prompt, target_audience, format, tone, active_tab],
-        outputs=[results],
-    )
-
+        city_input = gr.Dropdown(choices=cities.get("cities"), label="Select City")
+        industry_input = gr.Dropdown(choices=industries.get("industries"), label="Select Industry")
+        submit_button = gr.Button("Generate Report")
+
+    gr.Markdown("---")
+    gr.Markdown("---")
+    gr.Markdown("---")
+
+    report_output = gr.Markdown(value="", label="Report", visible=False)
+    
+    def on_submit(city, industry):
+        return gr.update(value=generate_report(city, industry), visible=True)
+    
+    submit_button.click(on_submit, inputs=[city_input, industry_input], outputs=report_output)
 
 def launch():
     """

src/report_genie/config/agents.yaml

@@ -1,10 +1,30 @@
-content_creator:
+data_researcher:
   role: >
-    Senior Content Creator
+    Data Researcher
   goal: >
-    Created engaging and informative content
+    Gather comprehensive information about specific companies that are in relevant cities and industries
   backstory: >
-    You're a seasoned content creator with a knack for producing high-quality
-    content that captivates and educates your audience. You're known for your
-    ability to translate complex concepts into clear and engaging narratives,
-    making it easy for others to learn and understand.
+    You are an expert data researcher with deep knowledge of
+    business ecosystems and city demographics. You excel at analyzing
+    complex data relationships.
+
+news_analyst:
+  role: >
+    News Analyst
+  goal: >
+    Find and analyze recent news about relevant companies in the specified industry and city
+  backstory: >
+    You are a seasoned news analyst with expertise in
+    business journalism and market research. You can identify key trends
+    and developments from news articles.
+
+report_writer:
+  role: >
+    Report Writer
+  goal: >
+    Create comprehensive, well-structured reports combining the provided research and news analysis. Do not include any information that isnt explicitly provided.
+  backstory: >
+    You are a professional report writer with experience in
+    business intelligence and market analysis. You excel at synthesizing
+    information into clear, actionable insights. Do not include any information that isn't explicitly provided.
+    

src/report_genie/config/tasks.yaml

@@ -1,40 +1,25 @@
-content_creator_task:
+city_research_task:
   description: >
-    Understand the context carefully
-    
-    ---
-    Context:
-    ({context})
-    ---
-
-    Describe the context in a way that would follow the tone, format, and guidelines provided.
-
-    ---
-    Tone:(
-    [Tone: {tone[name]}] ###  
-    [Tone Description: {tone[description]}] ### 
-    )
-    ---
-
-    ---
-    Format:(
-    [Format Type: {format[name]}] ### 
-    [Format Description:{format[description]}] ### 
-    [Max Content Length: {format[max_length]}] ### 
-    [Negetive: {format[negative]}])
-    )
-    ---
+    Research and analyze {city_name} and its business ecosystem in {industry_name} industry:
+        1. Get city summary and key information
+        2. Find organizations in the specified industry
+        3. Analyze business relationships and economic indicators
+  expected_output: >
+    Basic statistics about the companies in the given city and industry as well as top performers
+  agent: data_researcher
 
-    ---
-    Target Audience: (
-    [Target Audience: {target_audience[name]}] ### 
-    [Target Audience Description: {target_audience[description]}] ### 
-    [Ideal Audience Description: {target_audience[ideal_audience]}] ### 
-    )
-    ---
+news_analysis_task:
+  description: >
+    Analyze recent news about the companies provided by the city researcher
+  expected_output: >
+    Summarization of the latest news for the company and how it might affect the market
+  agent: news_analyst
 
-    Make sure you find the latest information about the topic in the internet if needed and provide a well-researched content.
+report_writing_task:
+  description: >
+    Create a detailed markdown report about the
+    results you got from city research and news analysis tasks.
+    Do not include any information that isn't provided
   expected_output: >
-    A fully fledge output with the desised format & tone and the the Guidelines provided in the description.
-    Formatted as markdown without '```'
-  agent: content_creator
+    Markdown summary report with key insights and recommendations.
+  agent: report_writer

src/report_genie/crew.py

@@ -7,6 +7,7 @@ import os
 from typing import Dict, Any, Optional
 import json
 from report_genie.utils.utils import load_json_data, sanitize_input
+from report_genie.tools.neo4j_tools import get_city_info, get_news
 
 load_dotenv()
 
@@ -21,134 +22,92 @@ KNOWLEDGE_SOURCE_PATH = "knowledge"
 
 
 @CrewBase
-class ExpresslyServer:
+class ReportGenieServer:
     """ExpresslyServer crew"""
 
-    # Create a knowledge source
-    json_knowledge_source = JSONKnowledgeSource(
-        file_paths=["format.json", "tone.json", "target_audience.json"],
-    )
-
     agents_config = "config/agents.yaml"
     tasks_config = "config/tasks.yaml"
 
     llm = LLM(model=MODEL, api_key=GEMINI_API_KEY, temperature=0.7)
 
-    @before_kickoff
-    def validate_inputs(
-        self, inputs: Optional[Dict[str, Any]]
-    ) -> Optional[Dict[str, Any]]:
+    @agent
+    def data_researcher(self) -> Agent:
         """
-        Validate and process user inputs based on the active tab selection.
-
-        This method checks the integrity and presence of required inputs, loads
-        necessary JSON data, and validates the active_tab value to ensure the
-        appropriate fields are populated. It formats the inputs for further processing.
-
-        Parameters:
-        inputs (Optional[Dict[str, Any]]): The dictionary containing user inputs,
-        including 'target_audience', 'format', 'tone', 'active_tab', and 'prompt'.
-
-        Returns:
-        Optional[Dict[str, Any]]: A dictionary formatted with context, format, tone,
-        and target_audience details based on the inputs provided.
-
-        Raises:
-        ValueError: If inputs are missing, not a dictionary, or required fields
-        ('active_tab', 'prompt', 'format', 'tone', 'target_audience') are not provided
-        or invalid.
+        The data_researcher agent is responsible for fetching data from the given data sources.
+        It takes the data sources as input and returns the data as JSON.
         """
-
-        if inputs is None or len(inputs) == 0 or not isinstance(inputs, dict):
-            raise ValueError("Inputs is required and must be a dictionary")
-
-        ## Get the first element from the list of inputs and get the value of target, format, active_tab and prompt
-        query = inputs
-        target_audience: str = sanitize_input(query.get("target_audience"))
-        content_format: str = sanitize_input(query.get("format"))
-        content_tone: str = sanitize_input(query.get("tone"))
-        prompt: str = query.get("prompt")
-
-        # Check if prompt are not None
-        if prompt is None:
-            raise ValueError("Prompt is required")
-
-        # Load JSON data from content_style_mapping.json
-        content_style_mapping_json = load_json_data(
-            CONTENT_STYLE_MAPPING_JSON_FILE, KNOWLEDGE_SOURCE_PATH
-        )
-        tone_json = load_json_data(TONE_JSON_FILE, KNOWLEDGE_SOURCE_PATH)
-        format_json = load_json_data(FORMAT_JSON_FILE, KNOWLEDGE_SOURCE_PATH)
-        target_audience_json = load_json_data(
-            TARGET_AUDIENCE_JSON_FILE, KNOWLEDGE_SOURCE_PATH
+        return Agent(
+            config=self.agents_config["data_researcher"],
+            llm=self.llm,
+            tools=[get_city_info],
+            verbose=True,
         )
 
-        if target_audience != "":
-            ## Resetting the format and tone as per the target audience
-            mappings: dict = content_style_mapping_json.get("target_audience").get(
-                target_audience
-            )
-            format_dict = format_json.get("format").get(mappings.get("format"))
-            tone_dict = tone_json.get("tone").get(mappings.get("tone"))
-            target_audience_dict = target_audience_json.get("target_audience").get(
-                target_audience
-            )
-        elif content_format != "" and content_tone != "":
-            ## Constructing a target_audience_dict with empty values and populating the format and tone as per the input
-            target_audience_dict = {
-                "name": "",
-                "description": "",
-                "ideal_audience": "",
-            }
-            format_dict = format_json.get("format").get(content_format)
-            tone_dict = tone_json.get("tone").get(content_tone)
-        else:
-            raise ValueError("Provide either target audience or format and tone")
-
-        ## Format the inputs
-        inputs = {
-            "context": prompt,
-            "format": format_dict,
-            "tone": tone_dict,
-            "target_audience": target_audience_dict,
-        }
-
-        return inputs
-
     @agent
-    def content_creator(self) -> Agent:
+    def news_analyst(self) -> Agent:
         """
-        Initializes and returns an Agent for content creation.
-
-        This agent is configured using predefined settings for the content creator
-        and utilizes a language model (LLM) for generating content. The agent
-        accesses a JSON knowledge source to enhance its capabilities and operates
-        in verbose mode for detailed output logging.
-
-        Returns:
-            Agent: An initialized agent configured for content creation.
+        The news_analyst agent is responsible for analyzing news articles
+        related to the data fetched by the data_researcher agent.
+        It takes the data sources as input and returns a JSON object
+        containing the analysis results.
         """
+        return Agent(
+            config=self.agents_config["news_analyst"],
+            llm=self.llm,
+            tools=[get_news],
+            verbose=True,
+        )
 
+    @agent
+    def report_writer(self) -> Agent:
+        """
+        The report_writer agent is responsible for generating a report given the
+        results from the data_researcher and news_analyst agents.
+        It takes the results from the agents as input and returns a JSON object
+        containing the report.
+        """
         return Agent(
-            config=self.agents_config["content_creator"],
+            config=self.agents_config["report_writer"],
             llm=self.llm,
-            knowledge_source=[self.json_knowledge_source],
             verbose=True,
         )
 
     @task
-    def content_creator_task(self) -> Task:
+    def city_research_task(self) -> Task:
+        """
+        The city_research_task task is responsible for executing the data_researcher
+        and news_analyst agents in order.
+        It takes no input and returns a JSON object containing the results of the
+        agents.
         """
-        Initializes and returns a Task for content creation.
+        return Task(
+            config=self.tasks_config["city_research_task"],
+        )
 
-        This task is configured using predefined settings for content creation
-        and is used by the content creator agent to generate content.
+    @task
+    def news_analysis_task(self) -> Task:
+        """
+        The news_analysis_task task is responsible for executing the news_analyst
+        agent to analyze news articles related to the data fetched by the
+        data_researcher agent.
+        It takes no input and returns a JSON object containing the analysis results.
+        """
+        return Task(
+            config=self.tasks_config["news_analysis_task"],
+            context=[self.city_research_task()],
+        )
 
-        Returns:
-            Task: An initialized task configured for content creation.
+    @task
+    def report_writing_task(self) -> Task:
         """
+        The report_writing_task task is responsible for executing the report_writer
+        agent to generate a report based on the findings from previous tasks.
+        It takes no input and returns a JSON object containing the generated report.
+        """
+
         return Task(
-            config=self.tasks_config["content_creator_task"],
+            config=self.tasks_config["report_writing_task"],
+            context=[self.city_research_task(), self.news_analysis_task()],
         )
 
     @crew
@@ -159,6 +118,5 @@ class ExpresslyServer:
             agents=self.agents,
             tasks=self.tasks,
             process=Process.sequential,
-            knowledge_source=[self.json_knowledge_source],
             verbose=True,
         )

src/report_genie/main.py

@@ -2,7 +2,7 @@
 import sys
 import warnings
 
-from report_genie.crew import ExpresslyServer
+from report_genie.crew import ReportGenieServer
 import dotenv
 
 warnings.filterwarnings("ignore", category=SyntaxWarning, module="pysbd")
@@ -16,13 +16,11 @@ def run():
     """
 
     inputs = {
-        "prompt": "I want to thanks DeepLearning and John from the crewAI for this amazing course..",
-        "format": "Email",
-        "tone": "Friendly",
-        "target_audience": "",
+        "city_name": "Seattle",
+        "industry_name": "Hardware Companies",
     }
 
-    ExpresslyServer().crew().kickoff(inputs=inputs)
+    ReportGenieServer().crew().kickoff(inputs=inputs)
 
 
 def train():
@@ -31,7 +29,7 @@ def train():
     """
     inputs = {"topic": "AI LLMs"}
     try:
-        ExpresslyServer().crew().train(
+        ReportGenieServer().crew().train(
             n_iterations=int(sys.argv[1]), filename=sys.argv[2], inputs=inputs
         )
 
@@ -44,7 +42,7 @@ def replay():
     Replay the crew execution from a specific task.
     """
     try:
-        ExpresslyServer().crew().replay(task_id=sys.argv[1])
+        ReportGenieServer().crew().replay(task_id=sys.argv[1])
 
     except Exception as e:
         raise Exception(f"An error occurred while replaying the crew: {e}")
@@ -56,7 +54,7 @@ def test():
     """
     inputs = {"topic": "AI LLMs"}
     try:
-        ExpresslyServer().crew().test(
+        ReportGenieServer().crew().test(
             n_iterations=int(sys.argv[1]), openai_model_name=sys.argv[2], inputs=inputs
         )
 

src/report_genie/schemas/schema.py

@@ -20,3 +20,8 @@ class ChatInput(BaseModel):
 
 class ChatOutput(BaseModel):
     result: str = Field(..., description="The transformed text response")
+
+class GetCityInfoInput(BaseModel):
+    """Input schema for MyCustomTool."""
+    city: str = Field(..., description="City name")
+    industry: str = Field(..., description="Industry name")

src/report_genie/tools/neo4j_tools.py

@@ -0,0 +1,59 @@
+from crewai.tools import tool
+from neo4j import GraphDatabase
+import os
+
+## Neo4j connection details
+NEO_DB_URI = os.getenv("NEO_DB_URI")
+NEO_DB_USERNAME = os.getenv("NEO_DB_USERNAME")
+NEO_DB_PWD = os.getenv("NEO_DB_USERNAME")
+NEO_DB_DATABSE = os.getenv("NEO_DB_DATABSE")
+
+AUTH = (NEO_DB_USERNAME, NEO_DB_PWD)
+driver = GraphDatabase.driver(NEO_DB_URI, auth=AUTH)
+
+@tool("get_city_info")
+def get_city_info(city_name: str, industry_name: str) -> list:
+        """
+        Get various information about a city and industry
+
+        Given a city and an industry, fetch the following information from the graph database:
+        - The number of organizations in the given industry that are in the given city
+        - The number of public companies in the given industry that are in the given city
+        - The total revenue of all companies in the given industry that are in the given city
+        - The 5 companies in the given industry that are in the given city with the most employees
+
+        :param city_name: The name of the city
+        :param industry_name: The name of the industry
+        :return: A list of dictionaries, each containing the above information
+        """
+        data, _, _ = driver.execute_query("""MATCH (c:City)<-[:IN_CITY]-(o:Organization)-[:HAS_CATEGORY]->(i:IndustryCategory)
+                    WHERE c.name = 
+                    industry
+                    WITH o
+                    ORDER BY o.nbrEmployees DESC
+                    RETURN count(o) AS organizationCount,
+                    sum(CASE WHEN o.isPublic THEN 1 ELSE 0 END) AS publicCompanies,
+                    sum(o.revenue) AS combinedRevenue,
+                    collect(CASE WHEN o.nbrEmployees IS NOT NULL THEN o END)[..5] AS topFiveOrganizations""", city=city_name, industry=industry_name)
+        return [el.data() for el in data]
+
+
+@tool("get_news")
+def get_news(company: str) -> list:
+        """
+        Get the 5 most recent news articles mentioning a given company.
+
+        :param company: The name of the company to search for
+        :return: A list of dictionaries with the following keys:
+            - title: The title of the article
+            - date: The date the article was published
+            - sentiment: The sentiment of the article (positive, negative, or neutral)
+            - chunks: A list of strings containing the text of the article
+        """
+        data, _, _ = driver.execute_query("""MATCH (c:Chunk)<-[:HAS_CHUNK]-(a:Article)-[:MENTIONS]->(o:Organization)
+                    WHERE o.name = $company AND a.date IS NOT NULL
+                    WITH c, a
+                    ORDER BY a.date DESC
+                    LIMIT 5
+                    RETURN a.title AS title, a.date AS date, a.sentiment AS sentiment, collect(c.text) AS chunks""", company=company)
+        return [el.data() for el in data]

src/report_genie/utils/utils.py

@@ -14,7 +14,7 @@ def load_json_data(file_name: str, file_path: str) -> dict:
     """
 
     file_path = f"{file_path}/{file_name}"
-    with open(file_path, "r") as file:
+    with open(file_path, "r",  encoding="utf-8-sig") as file:
         return json.load(file)
 
 

uv.lock

@@ -2318,6 +2318,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/2a/e2/5d3f6ada4297caebe1a2add3b126fe800c96f56dbe5d1988a2cbe0b267aa/mypy_extensions-1.0.0-py3-none-any.whl", hash = "sha256:4392f6c0eb8a5668a69e23d168ffa70f0be9ccfd32b5cc2d26a34ae5b844552d", size = 4695 },
 ]
 
+[[package]]
+name = "neo4j"
+version = "5.28.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "pytz" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/4b/20/733dac16f7cedc80b23093415822c9763302519cba0e7c8bcdb5c01fc512/neo4j-5.28.1.tar.gz", hash = "sha256:ae8e37a1d895099062c75bc359b2cce62099baac7be768d0eba7180c1298e214", size = 231094 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/6a/57/94225fe5e9dabdc0ff60c88cbfcedf11277f4b34e7ab1373d3e62dbdd207/neo4j-5.28.1-py3-none-any.whl", hash = "sha256:6755ef9e5f4e14b403aef1138fb6315b120631a0075c138b5ddb2a06b87b09fd", size = 312258 },
+]
+
 [[package]]
 name = "networkx"
 version = "3.4.2"
@@ -3619,12 +3631,14 @@ source = { editable = "." }
 dependencies = [
     { name = "crewai", extra = ["tools"] },
     { name = "gradio" },
+    { name = "neo4j" },
 ]
 
 [package.metadata]
 requires-dist = [
     { name = "crewai", extras = ["tools"], specifier = ">=0.86.0,<1.0.0" },
     { name = "gradio", specifier = ">=5.12.0,<5.13.0" },
+    { name = "neo4j", specifier = ">=5.28.1,<6.0.0" },
 ]
 
 [[package]]