AI Threats

# ===== Fixed Parameters =====

# Keywords for filtering content
keywords = ['AI']

# Risk Theme
main_theme_risk = 'Risks or concerns related to AI disruption'

# Proactivity Theme
main_theme_proactivity = 'Integrating GenAI into products or forming partnerships to improve GenAI offerings'

# Sentences expressing risk-related narratives
list_sentences_risks = [
    "Companies face substantial risks from AI-driven market disruptions today.",
    "AI advancements threaten traditional business models and competitive advantages.",
    "Disruption by AI can lead to loss of market share rapidly.",
    "Existing workflows may become obsolete due to AI innovations.",
    "Companies risk being outpaced by agile AI-focused competitors."
]

# Sentences expressing proactivity-related narratives
list_sentences_proactivity = [
    "Companies actively incorporate GenAI to enhance product offerings and features.",
    "Partnerships aim to improve GenAI solutions in competitive markets.",
    "Proactive firms leverage GenAI technology to boost customer engagement strategies.",
    "Organizations collaborate to advance GenAI capabilities and applications effectively.",
    "Firms invest in GenAI partnerships to innovate their service delivery."
]

# ===== Customizable Parameters =====

# Company Universe (from Watchlist)
my_watchlist_id = "37fe5403-9688-4313-9775-9fd9770ca6e2"
watchlist = bigdata.watchlists.get(my_watchlist_id)
companies = bigdata.knowledge_graph.get_entities(watchlist.items)

# LLM Specification
llm_model = "openai::gpt-4o-mini"

# Search Frequency
search_frequency='M'

# Specify Time Range
start_date="2025-01-01"
end_date="2025-04-20"

# Fiscal Year
fiscal_year = 2025

# Document Limits
document_limit_news=10
document_limit_filings=5
document_limit_transcripts=5

# Others
batch_size=1

report_generator = GenerateReport(
        watchlist_id=my_watchlist_id,
        keywords=['AI'],
        main_theme_risk='Risks or concerns related to AI disruption',
        main_theme_proactivity='Integrating GenAI into products or forming partnerships to improve GenAI offerings',
        list_sentences_risks=list_sentences_risks,
        list_sentences_proactivity=list_sentences_proactivity,
        llm_model= llm_model,
        api_key=OPENAI_API_KEY,
        start_date=start_date,
        end_date=end_date,
        fiscal_year = fiscal_year,
        search_frequency=search_frequency,
        document_limit_news=document_limit_news,
        document_limit_filings=document_limit_filings,
        document_limit_transcripts=document_limit_transcripts,
        batch_size=batch_size,
        bigdata=bigdata
)

# Initialize empty lists to collect dataframes
df_risk_list = []
df_proactivity_list = []

# Define the document types and limits
doc_configs = [
   (DocumentType.NEWS, document_limit_news, None),
   (DocumentType.FILINGS, document_limit_filings, fiscal_year),
   (DocumentType.TRANSCRIPTS, document_limit_transcripts, fiscal_year)
]

# Iterate through each document type configuration
for doc_type, doc_limit, year in doc_configs:
   if doc_limit > 0:
       # Search for semantic risk sentences
       df_sentences_semantic_risk_search = search_by_companies(
           companies=companies,
           keywords=keywords,
           sentences=list_sentences_risks,
           fiscal_year=year,
           start_date=start_date,
           end_date=end_date,
           scope=doc_type,
           freq=search_frequency,
           document_limit=doc_limit,
           batch_size=batch_size
       )
       df_risk_list.append(df_sentences_semantic_risk_search)

       # Search for semantic proactivity sentences
       df_sentences_semantic_proactivity_search = search_by_companies(
           companies=companies,
           keywords=keywords,
           sentences=list_sentences_proactivity,
           fiscal_year=year,
           start_date=start_date,
           end_date=end_date,
           scope=doc_type,
           freq=search_frequency,
           document_limit=doc_limit,
           batch_size=batch_size
       )
       df_proactivity_list.append(df_sentences_semantic_proactivity_search)

# Combine all risk dataframes into a single dataset
df_sentences_semantic_risk = pd.concat(df_risk_list, ignore_index=True)

# Combine all proactivity dataframes into a single dataset
df_sentences_semantic_proactivity = pd.concat(df_proactivity_list, ignore_index=True)

labeler = ScreenerLabeler(llm_model=llm_model)

df_risk_labels = labeler.get_labels(
                main_theme=main_theme_risk,
                labels=['risk'],
                texts=df_sentences_semantic_risk["masked_text"].tolist()
            )

# Merge the datasets
df_risk_labeled = pd.merge(df_sentences_semantic_risk, df_risk_labels, left_index=True, right_index=True)

df_proactivity_labels = labeler.get_labels(
                main_theme=main_theme_proactivity,
                labels=['proactivity'],
                texts=df_sentences_semantic_risk["masked_text"].tolist()
            )


# Merge the datasets
df_proactivity_labeled = pd.merge(df_sentences_semantic_proactivity, df_proactivity_labels, left_index=True, right_index=True)

# Process the results
df_risk_labeled_relevant = df_risk_labeled.loc[~df_risk_labeled.label.isin(['', 'unassigned', 'unclear'])].copy()
df_proactivity_labeled_relevant = df_proactivity_labeled.loc[~df_proactivity_labeled.label.isin(['', 'unassigned', 'unclear'])].copy()

create_styled_table(df_risk_labeled_relevant, title='Relevant Document Count for AI Disruption Risk by Company and Document Type', companies_list = company_names)

create_styled_table(df_proactivity_labeled_relevant, title='Relevant Document Count for AI Proactivity by Company and Document Type', companies_list = company_names)

summarizer_company = SummarizerCompany(
    model=llm_model.split('::')[1],
    api_key=OPENAI_API_KEY,
    logger= GenerateReport.logger,
    verbose=True
)

df_risk_by_company = asyncio.run(
            summarizer_company.process_by_company(
                df_labeled=df_risk_labeled_relevant,
                list_entities=companies,
                theme=main_theme_risk,
                focus='risk'
            ))

df_proactivity_by_company = asyncio.run(
            summarizer_company.process_by_company(
                df_labeled=df_proactivity_labeled_relevant,
                list_entities=companies,
                theme=main_theme_proactivity,
                focus=''
            )
        )

dfr = df_risk_by_company[['entity_id', 'entity_name', 'topic_summary', 'n_documents']].copy()
dfr = dfr.rename(columns={'topic_summary': 'risk_summary', 'n_documents': 'n_documents_risk'})
dfp = df_proactivity_by_company[['entity_id', 'entity_name', 'topic_summary', 'n_documents']].copy()
dfp = dfp.rename(columns={'topic_summary': 'proactivity_summary', 'n_documents': 'n_documents_proactivity'})
df_by_company = dfr.merge(dfp, on=['entity_id', 'entity_name'], how='outer')

df_by_company['n_documents_risk'] = df_by_company['n_documents_risk'].fillna(0)
df_by_company['n_documents_proactivity'] = df_by_company['n_documents_proactivity'].fillna(0)
df_by_company['ai_disruption_risk_score'] = df_by_company['n_documents_risk']/df_by_company['n_documents_risk'].mean()
df_by_company['ai_proactivity_score'] = df_by_company['n_documents_proactivity']/df_by_company['n_documents_proactivity'].mean()
df_by_company['ai_proactivity_minus_disruption_risk_score'] = df_by_company['ai_proactivity_score'] - df_by_company['ai_disruption_risk_score']

df_quotes_risk = GenerateReport.aggregate_verbatim(df_risk_labeled_relevant, 'risk')
df_by_company = pd.merge(df_by_company, df_quotes_risk, how='left', on='entity_name')
df_quotes_proactivity = GenerateReport.aggregate_verbatim(df_proactivity_labeled_relevant, 'proactivity')
df_report = pd.merge(df_by_company, df_quotes_proactivity, how='left', on='entity_name')

plot_company_scores_log_log(df_report, 'ai_disruption_risk_score', 'ai_proactivity_score', 'AI Narrative Signals')

plot_company_scores(df_report, 'ai_disruption_risk_score', 'ai_proactivity_score', 'AI Narrative Signals')

display_report(df_report=df_report, score='ai_proactivity_minus_disruption_risk_score', top='top', nb_entities=5, export_to_path=output_dir+'/report_ai_disruption_top_5.html')

try:
    # Create the Excel manager
    excel_manager = ExcelManager()

    # Define the dataframes and their sheet configurations
    df_args = [
        (df_report, "Report AI Disruption Risk", (2, 3))
    ]

    # Save the workbook
    excel_manager.save_workbook(df_args, export_path)

except Exception as e:
    print(f"Warning while exporting to excel: {e}")