• Automated Company Research — Enter a company name and get a structured report covering leadership, products, financials, and recent news. The system pulls information from the company website and multiple web searches, then synthesizes everything into a ready-to-use analysis.
• Email-Triggered Research — Send a research request by email with just a company name or description. The system fills in missing details automatically, runs the full research pipeline, and replies with the finished report directly to your inbox.

• Nested Aktor Compositions — aktorCompanyResearch is a defineAktor subgraph that orchestrates scraping, search, and AI analysis. Inside it, aktorAIAnalyzeCompany is another nested defineAktor handling prompt construction and the LLM call. This multi-level composition keeps each concern isolated while building complex pipelines.
• Stream-Based Search Processing — Company research generates four search queries via aktorSearchCompanyInfo. aktorForEachStream creates a stream, aktorInput holds the current query, and aktorStreamProcessor runs aktorWebsearch for each item. Results are combined and formatted before AI analysis.
• Web Scraping via Jina AI — aktorWebscrape fetches the company website as markdown with extracted links. The scraped content is processed and structured via aktorProcessCompanyData before being passed alongside search results to the AI for analysis.
• Tool Wrapping for Email Automation — The mail Reaktor uses aktorToTool to expose both aktorCompanyResearch and aktorWebsearch as AI-callable tools, combined via aktorToolSet. The AI autonomously decides: search first to find missing details, then invoke company research when it has enough data — up to five tool iterations.
• Configurable Prompt Templates — All system and user prompts use aktorCompletePrompt with {{variable}} placeholders. Prompts are exposed as aktorSetting values, so research instructions, output structure, and language can be customized through the Operaide UI without code changes.

The Company Research app registers two Reaktors. The company research Reaktor is the core: it scrapes the target website via aktorWebscrape, processes the content with aktorProcessCompanyData, generates four search queries via aktorSearchCompanyInfo, and processes them through aktorStreamProcessor with aktorWebsearch. The formatted search results and scraped website data flow into a nested aktorAIAnalyzeCompany subgraph, where an AI produces the final structured report. The mail Reaktor wraps company research and web search as AI-callable tools via aktorToTool and aktorToolSet, lets the AI decide which tools to use based on the email content, and sends the result back via aktorReplyToMail.