Add brand monitoring code

Author: ChawlaAvi

brand-monitoring/README.md

@@ -0,0 +1,72 @@
+# Brand monitoring flow using DeepSeek-R1, CrewAI and BrightData
+
+This project implements an automated brand monitoring system using AI agents. We use the following tools to build this:
+- [Bright Data](https://brdta.com/dailydoseofds) is used to scrape the web.
+- CrewAI to build the Agentic workflow.
+- DeepSeek-R1 as the LLM.
+
+The brand monitoring output is shown here: [Sample output](brand-monitoring-demo.mp4)
+
+---
+## Setup and installations
+
+**Get BrightData API Key**:
+- Go to [Bright Data](https://brdta.com/dailydoseofds) and sign up for an account.
+- Select "Proxies & Scraping" and create a new "SERP API"
+- Select "Native proxy-based access"
+- You will find your username and password there.
+- Store it in the .env file of the src/ folder (after renaming the .env.example file to .env)
+
+```
+BRIGHT_DATA_USERNAME="..."
+BRIGHT_DATA_PASSWORD="..."
+```
+
+- Also get the Bright Data API key from your dashboard.
+
+```
+BRIGHT_DATA_API_KEY="..."
+```
+
+**Setup Ollama**:
+   ```bash
+   # setup ollama on linux 
+   curl -fsSL https://ollama.com/install.sh | sh
+   # pull deepseek-r1 model
+   ollama pull deepseek-r1
+   ```
+
+
+**Install Dependencies**:
+   Ensure you have Python 3.11 or later installed.
+   ```bash
+   pip install ollama crewai crewai-tools streamlit
+   ```
+
+---
+
+## Run the project
+
+Finally, head over to this folder:
+```
+cd brand_monitoring_flow/src
+```
+
+and run the project by running the following command:
+
+```bash
+streamlit run brand_monitoring_app.py
+```
+
+---
+
+## 📬 Stay Updated with Our Newsletter!
+**Get a FREE Data Science eBook** 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. [Subscribe now!](https://join.dailydoseofds.com)
+
+[![Daily Dose of Data Science Newsletter](https://github.com/patchy631/ai-engineering/blob/main/resources/join_ddods.png)](https://join.dailydoseofds.com)
+
+---
+
+## Contribution
+
+Contributions are welcome! Please fork the repository and submit a pull request with your improvements.

brand-monitoring/brand-monitoring-demo.mp4

@@ -0,0 +1,3 @@
+BRIGHT_DATA_USERNAME="..."
+BRIGHT_DATA_PASSWORD="..."
+BRIGHT_DATA_API_KEY="..."

brand-monitoring/brand_monitoring_flow/.env.example

@@ -0,0 +1,20 @@
+[project]
+name = "brand_monitoring_flow"
+version = "0.1.0"
+description = "brand-monitoring-flow using crewAI"
+authors = [{ name = "Your Name", email = "you@example.com" }]
+requires-python = ">=3.10,<3.13"
+dependencies = [
+    "crewai[tools]>=0.102.0,<1.0.0",
+]
+
+[project.scripts]
+kickoff = "brand_monitoring_flow.main:kickoff"
+plot = "brand_monitoring_flow.main:plot"
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.crewai]
+type = "flow"

brand-monitoring/brand_monitoring_flow/pyproject.toml

@@ -0,0 +1,159 @@
+import streamlit as st
+import base64
+import gc
+import brand_monitoring_flow.main
+
+# ===========================
+#   Streamlit Setup
+# ===========================
+
+if "response" not in st.session_state:
+    st.session_state.response = None
+
+if "flow" not in st.session_state:
+    st.session_state.flow = None
+
+if "deep_seek_image" not in st.session_state:
+    st.session_state.deep_seek_image = base64.b64encode(open("assets/deep-seek.png", "rb").read()).decode()
+if "brightdata_image" not in st.session_state:
+    st.session_state.brightdata_image = base64.b64encode(open("assets/brightdata.png", "rb").read()).decode()
+
+def reset_analysis():
+    st.session_state.response = None
+    st.session_state.flow = None
+    gc.collect()
+
+
+def start_analysis():
+    st.markdown("""
+            # Brand Monitoring powered by <img src="data:image/png;base64,{}" width="180" style="vertical-align: -7px;"> & <img src="data:image/png;base64,{}" width="180" style="vertical-align: -10px;">
+        """.format(
+            st.session_state.deep_seek_image,
+            st.session_state.brightdata_image
+        ), unsafe_allow_html=True)
+    
+    # Create a placeholder for status updates
+    status_placeholder = st.empty()
+    
+    with status_placeholder.container():
+        if st.session_state.flow is None:
+            status_placeholder.info('Initializing brand monitoring flow...')
+            st.session_state.flow = brand_monitoring_flow.main.BrandMonitoringFlow()
+        
+            st.session_state.flow.state.brand_name = st.session_state.brand_name
+            st.session_state.flow.state.total_results = st.session_state.total_results
+            
+            # You can update the status for different phases
+            status_placeholder.info('Scraping latest data about {} and analyzing with AI Agents...'.format(st.session_state.brand_name))
+            st.session_state.flow.kickoff()
+            
+            # Store the results
+            status_placeholder.success('Analysis complete! Displaying results...')
+            st.session_state.response = st.session_state.flow.state
+        else:
+            st.session_state.response = st.session_state.flow.state
+
+    # Clear the status message after completion
+    status_placeholder.empty()
+
+# ===========================
+#   Sidebar
+# ===========================
+with st.sidebar:
+    st.header("Brand Monitoring Settings")
+    
+    # Brand name input
+    st.session_state.brand_name = st.text_input(
+        "Company/Brand Name",
+        value="Hugging Face" if "brand_name" not in st.session_state else st.session_state.brand_name
+    )
+    
+    # Number of search results
+    st.session_state.total_results = st.number_input(
+        "Total Search Results",
+        min_value=1,
+        max_value=50,
+        value=15,
+        step=1
+    )
+
+    st.divider()
+    
+    # Analysis buttons
+    col1, col2 = st.columns(2)
+    with col1:
+        st.button("Start Analysis 🚀", type="primary", on_click=start_analysis)
+    with col2:
+        st.button("Reset", on_click=reset_analysis)
+
+# ===========================
+#   Main Content Area
+# ===========================
+
+# Move the header inside a container to ensure it stays at the top
+
+if st.session_state.response is None:
+    header_container = st.container()
+    with header_container:
+        st.markdown("""
+            # Brand Monitoring powered by <img src="data:image/png;base64,{}" width="180" style="vertical-align: -7px;"> & <img src="data:image/png;base64,{}" width="180" style="vertical-align: -10px;">
+        """.format(
+            st.session_state.deep_seek_image,
+            st.session_state.brightdata_image
+        ), unsafe_allow_html=True)
+
+if st.session_state.response:
+    try:
+        response = st.session_state.response
+        
+        # LinkedIn Results
+        if response.linkedin_crew_response:
+            st.markdown("## 💼 LinkedIn Mentions")
+            for post in response.linkedin_crew_response.pydantic.content:
+                with st.expander(f"📝 {post.post_title}"):
+                    st.markdown(f"**Source:** [{post.post_link}]({post.post_link})")
+                    for line in post.content_lines:
+                        st.markdown(f"- {line}")
+        
+        # Instagram Results
+        if response.instagram_crew_response:
+            st.markdown("## 📸 Instagram Mentions")
+            for post in response.instagram_crew_response.pydantic.content:
+                with st.expander(f"📝 {post.post_title}"):
+                    st.markdown(f"**Source:** [{post.post_link}]({post.post_link})")
+                    for line in post.content_lines:
+                        st.markdown(f"- {line}")
+        
+        # YouTube Results
+        if response.youtube_crew_response:
+            st.markdown("## 🎥 YouTube Mentions")
+            for video in response.youtube_crew_response.pydantic.content:
+                with st.expander(f"📝 {video.video_title}"):
+                    st.markdown(f"**Source:** [{video.video_link}]({video.video_link})")
+                    for line in video.content_lines:
+                        st.markdown(f"- {line}")
+        
+        # X/Twitter Results
+        if response.x_crew_response:
+            st.markdown("## 🐦 X/Twitter Mentions")
+            for post in response.x_crew_response.pydantic.content:
+                with st.expander(f"📝 {post.post_title}"):
+                    st.markdown(f"**Source:** [{post.post_link}]({post.post_link})")
+                    for line in post.content_lines:
+                        st.markdown(f"- {line}")
+        
+        # Web Results
+        if response.web_crew_response:
+            st.markdown("## 🌐 Web Mentions")
+            for page in response.web_crew_response.pydantic.content:
+                with st.expander(f"📝 {page.page_title}"):
+                    st.markdown(f"**Source:** [{page.page_link}]({page.page_link})")
+                    for line in page.content_lines:
+                        st.markdown(f"- {line}")
+
+    except Exception as e:
+        st.error(f"An error occurred while displaying results: {str(e)}")
+
+# Footer
+st.markdown("---")
+st.markdown("Built with CrewAI, Bright Data and Streamlit") 

brand-monitoring/brand_monitoring_flow/src/assets/brightdata.png

@@ -0,0 +1,62 @@
+from crewai import Agent, Crew, Process, Task
+from crewai.project import CrewBase, agent, crew, task
+from pydantic import BaseModel, Field   
+
+class XWriterReport(BaseModel):
+    post_title: str = Field(description="The title explaining how the brand was used in an individual post")
+    post_link: str = Field(description="The link to the scraped content")
+    content_lines: list[str] = Field(description="The bullet points within the scraped content that are relevant to the brand")
+
+class XReport(BaseModel):   
+    content: list[XWriterReport] = Field(description=("A list of extracted content with title, the post link, "
+                                                              "and the bullet points within each unique post. "
+                                                              "The size of the output list will be the same as the number of posts in the input data.")
+                                                              )
+
+llm = LLM(model="ollama/deepseek-r1")
+
+
+@CrewBase
+class XCrew:
+    """X Analysis Crew"""
+
+    agents_config = "config/agents.yaml"
+    tasks_config = "config/tasks.yaml"
+
+    @agent
+    def analysis_agent(self) -> Agent:
+        return Agent(
+            config=self.agents_config["analysis_agent"],
+            llm=llm,
+        )
+        
+    @task
+    def analysis_task(self) -> Task:
+        return Task(
+            config=self.tasks_config["analysis_task"],
+        )
+    
+    @agent
+    def writer_agent(self) -> Agent:
+        return Agent(
+            config=self.agents_config["writer_agent"],
+            llm=llm,
+        )
+
+    @task
+    def write_report_task(self) -> Task:
+        return Task(
+            config=self.tasks_config["write_report_task"],
+            output_pydantic=XReport,
+        )
+
+    @crew
+    def crew(self) -> Crew:
+        """Creates the X Analysis Crew"""
+
+        return Crew(
+            agents=self.agents,
+            tasks=self.tasks,
+            process=Process.sequential,
+            verbose=True,
+        )