from typing import Type from crewai.tools import BaseTool from pydantic import BaseModel, Field import dspy import os from dotenv import load_dotenv load_dotenv(dotenv_path="/home/azureuser/microlearn/backend/prompt_opt_dspy/.env") azure_api_key = os.getenv("AZURE_API_KEY") azure_api_host = os.getenv("AZURE_API_HOST") azure_deployment_id = os.getenv("AZURE_DEPLOYMENT_ID") azure_api_version = os.getenv("AZURE_API_VERSION") # === Configure Azure OpenAI LM === lm = dspy.LM( "azure/Csqr-gpt-4o-mini", api_key=azure_api_key, api_base=azure_api_host, api_version=azure_api_version, temperature=0.0, ) dspy.configure(lm=lm) # === DSPy Signature === class ResearchSignature(dspy.Signature): """ You are a topic researcher for a user in a specific topic. Uncover cutting-edge developments about topic The user prefer the researched data to be in match with the company context. The company context is : Company Name : {Company_Name} Industry : {Industry} Company Size : {Company_Size} Business Model : {Business_Model} Top Use Cases : {Top_Use_Cases} Learning Focus : {Learning_Focus} Tech Stack : {Tech_stack} Compliance : {Compliance} Perform comprehensive research on the topic **{topic}**, keeping in mind the user's job title, skills, experience, and learning level the motive of the user to under go this research. Avoid going too deep into technical details — this content is intended for a non-technical or minimally technical audience seeking a well-rounded understanding of the topic. Identify and list relevant concepts, methods, tools, techniques, and subtopics that should be included in a structured user journey. Your research should cover essential fundamentals, applied tools, advanced topics (aligned with the {Level} level), and current trends relevant to the user's job title **{Job_Title}** and skills **{Skills}** and motive **{motive}**. A list of researched user journey content items for the topic {topic} tailored to the {Level} level. Structure the result as: - **Topic Area**: Main area/category (e.g., Fundamentals, Applications) - **Subtopics/Concepts**: List of individual concepts or tools under each topic area. This is purely research-based content gathering for a non-technical or minimally technical user journey preparation. """ topic: str = dspy.InputField() company_context: str = dspy.InputField() user_context: str = dspy.InputField() researched_content: str = dspy.OutputField( desc="Structured bullet list of topic areas and subtopics tailored to the user journey" ) # === DSPy Agent with Optimizer (2-shot training) === class ResearcherModule(dspy.Module): def __init__(self): super().__init__() predictor = dspy.Predict(ResearchSignature) # --- Two Example Training Shots --- trainset = [ dspy.Example( topic="AI in Healthcare", company_context=""" Company Name: MediCore Solutions Industry: Healthcare Company Size: 2000+ employees Business Model: B2B and B2C SaaS Top Use Cases: Predictive analytics, Patient monitoring Learning Focus: AI diagnostics and workflow automation Tech Stack: Python, PyTorch, AWS HealthLake Compliance: HIPAA, GDPR """, user_context=""" Job Title: Healthcare Analyst Skills: Data Analysis, Clinical Research Level: Intermediate motive: Exploring AI applications to improve patient care """, researched_content=""" - **Topic Area**: Fundamentals - **Subtopics/Concepts**: - Introduction to Artificial Intelligence in Healthcare - Overview of Machine Learning and its Role in Patient Care - Types of AI: Supervised, Unsupervised, and Reinforcement Learning - Ethical Considerations in AI Applications - Understanding Data Privacy: HIPAA and GDPR Compliance - **Topic Area**: Applications - **Subtopics/Concepts**: - AI in Predictive Analytics: Enhancing Patient Outcomes - AI in Patient Monitoring: Wearables and Remote Monitoring - Use of Natural Language Processing (NLP) in Clinical Documentation - Chatbots and Virtual Health Assistants for Patient Engagement - Remote Diagnosis and Telemedicine Solutions - **Topic Area**: Tools and Techniques - **Subtopics/Concepts**: - Overview of Data Analytics Tools: Python and PyTorch for healthcare - Using AWS HealthLake for Data Management in Healthcare - Data Visualization Tools for Presenting AI Insights - Techniques in Data Preprocessing and Feature Engineering - Implementing AI Models for Real-Time Patient Monitoring - **Topic Area**: Advanced Topics - **Subtopics/Concepts**: - Deep Learning Applications in Medical Imaging - AI-Driven Decision Support Systems for Healthcare Professionals - Integrating AI into Existing Clinical Workflows - Case Studies on AI Success Stories in Patient Care - Challenges and Limitations of AI in Healthcare Practice - **Topic Area**: Current Trends - **Subtopics/Concepts**: - Rise of Telehealth Solutions and AI Integration - Innovations in Wearable Health Technology and AI - The Role of AI in Addressing Pandemic Challenges - The Future of AI in Personalized Medicine - Collaborations Between Tech Companies and Healthcare Institutions This structured user journey content is designed to provide a comprehensive overview of AI in Healthcare, particularly aligned with the interests and learning objectives of a Healthcare Analyst looking to explore how AI can improve patient care. """ ).with_inputs("topic", "company_context", "user_context"), dspy.Example( topic="Blockchain in FinTech", company_context=""" Company Name: FinNext Global Industry: FinTech Company Size: 1000-1500 employees Business Model: B2B Top Use Cases: Cross-border payments, Fraud prevention Learning Focus: Blockchain adoption and market opportunities Tech Stack: Ethereum, Hyperledger, Azure Blockchain Service Compliance: KYC, AML """, user_context=""" Job Title: Business Development Executive Skills: Partnerships, Client Engagement Level: Beginner motive: Understanding blockchain for business strategy alignment """, researched_content=""" - **Topic Area**: Fundamentals - **Subtopics/Concepts**: - Introduction to Blockchain: Understanding what blockchain technology is and its significance in the FinTech landscape. - Key Terms: Decentralization, ledgers, nodes, smart contracts, tokens, hash functions. - Differences between Blockchain Types: Public vs. Private vs. Consortium blockchains. - Basic Mechanisms: How transactions are processed and validated in a blockchain network. - Overview of Cryptography: Basic principles of how cryptography secures blockchain solutions. - **Topic Area**: Applications in FinTech - **Subtopics/Concepts**: - Cross-border Payments: How blockchain facilitates faster, cheaper, and transparent international money transfers. - Fraud Prevention: Use of blockchain to enhance security and traceability of transactions. - Digital Identity Management: Evaluating blockchain's role in KYC and AML compliance processes for identity verification. - Smart Contracts: Understanding automated contract execution and its applications in financial services. - **Topic Area**: Tools & Platforms - **Subtopics/Concepts**: - Overview of Ethereum and Hyperledger: Key differences, use cases, and their relevance in FinTech. - Azure Blockchain Service: Brief introduction to Microsoft’s blockchain service and its business applications. - Wallet Solutions: Insight into digital wallets and their importance for businesses. - Integration with Existing Systems: Exploring middleware and API solutions to connect blockchain with legacy systems. - **Topic Area**: Market Opportunities - **Subtopics/Concepts**: - Trends in Blockchain Adoption: Current and future trends driving blockchain in FinTech, including regulatory factors. - Competitive Landscape: Overview of companies and startups utilizing blockchain for competitive advantage. - Customer Engagement: How blockchain can improve client engagement through transparency and trust. - **Topic Area**: Business Strategy Alignment - **Subtopics/Concepts**: - Strategic Partnerships: Identifying potential partnerships in the blockchain ecosystem for business growth. - Business Models leveraging Blockchain: Understanding various B2B models, including payment solutions, lending, and insurance. - Measuring ROI: Exploring metrics and frameworks to assess the financial impact of blockchain initiatives. - **Topic Area**: Future Developments - **Subtopics/Concepts**: - Emerging Trends: Overview of upcoming blockchain technology trends such as NFTs, CBDCs (Central Bank Digital Currencies), and DeFi (Decentralized Finance). - Regulatory Landscape: Understanding evolving compliance standards and regulations impacting blockchain implementations. - Education and Skill Development: Resources and communities to explore for continuous learning in blockchain and FinTech. This structured user journey provides the necessary foundation and insight for a Business Development Executive at FinNext Global to understand blockchain's application in developing business strategies and enhancing client engagement in the FinTech sector. """ ).with_inputs("topic", "company_context", "user_context"), ] # Metric: Simple keyword containment def research_metric(gold, pred, trace): return int(all( section.lower().split()[0] in pred.researched_content.lower() for section in ["fundamentals", "applications", "tools", "trends"] )) # Compile with 2-shot training optimizer = dspy.BootstrapFewShot(metric=research_metric) optimizer.compile(predictor, trainset=trainset) # Store optimized predictor self.research_predictor = predictor print("=== Optimized ===") print(predictor) def forward(self, topic, company_context, user_context): return self.research_predictor( topic=topic, company_context=company_context, user_context=user_context ) # === Tool Input Schema === class ResearchToolInput(BaseModel): """Input schema for ResearchAnalysisTool.""" topic: str = Field(..., description="The topic to research") Company_Name: str = Field(..., description="Name of the company") Industry: str = Field(..., description="Industry of the company") Company_Size: str = Field(..., description="Size of the company") Business_Model: str = Field(..., description="Company's business model") Top_Use_Cases: str = Field(..., description="Top use cases the company focuses on") Learning_Focus: str = Field(..., description="Learning focus for the user") Tech_stack: str = Field(..., description="Company's tech stack") Compliance: str = Field(..., description="Relevant compliance requirements") Job_Title: str = Field(..., description="Job title of the user") Skills: str = Field(..., description="Skills of the user") Level: str = Field(..., description="Learning level (Beginner, Intermediate, Advanced)") motive: str = Field(..., description="The user's motive for research") # === Custom DSPy Tool === class ResearchAnalysisTool(BaseTool): name: str = "Research Analysis Tool" description: str = ( "Performs comprehensive research on a given topic while keeping in mind company context " "and user profile (job title, skills, level, and motive). Produces structured user journey " "content with topic areas and subtopics." ) args_schema: Type[BaseModel] = ResearchToolInput def _run( self, topic: str, Company_Name: str, Industry: str, Company_Size: str, Business_Model: str, Top_Use_Cases: str, Learning_Focus: str, Tech_stack: str, Compliance: str, Job_Title: str, Skills: str, Level: str, motive: str ) -> str: # Build company + user context company_context = f""" Company Name: {Company_Name} Industry: {Industry} Company Size: {Company_Size} Business Model: {Business_Model} Top Use Cases: {Top_Use_Cases} Learning Focus: {Learning_Focus} Tech Stack: {Tech_stack} Compliance: {Compliance} """ user_context = f""" Job Title: {Job_Title} Skills: {Skills} Level: {Level} Motive: {motive} """ researcher = ResearcherModule() result = researcher.forward(topic, company_context, user_context) return result.researched_content # ## Standalone Usage lines # research_tool = ResearchAnalysisTool() # # Run it # result = research_tool._run( # topic="Generative AI", # Company_Name="TechNova", # Industry="Healthcare", # Company_Size="5000 employees", # Business_Model="B2B SaaS", # Top_Use_Cases="Clinical decision support, patient engagement", # Learning_Focus="AI for non-technical staff", # Tech_stack="Python, TensorFlow, Azure Cloud", # Compliance="HIPAA, GDPR", # Job_Title="Product Manager", # Skills="Healthcare domain knowledge, basic analytics", # Level="Beginner", # motive="Understand AI use cases to build product strategy" # ) # print(result)