AI ThoughtStream

This application enables end-to-end AI-driven content generation and seamless publishing on LinkedIn. The workflow consists of intelligent agents collaborating to produce high-quality articles in the field of Data Science, Artificial Intelligence, and Machine Learning, complete with visualizations and social-ready summaries. Key Workflow Stages Topic Identification If the user hasn't specified a topic, the app generates a trending AI/ML topic using a topic scout agent. Otherwise, it uses the user-defined custom topic. Research & Summarization A research agent gathers key facts, statistics, and context from reliable AI sources. The summary includes 3–5 well-structured insights. Technical Article Writing A writing agent composes a clear, engaging article (~300–800 words) from the research. The content includes an intro, detailed body sections, and a conclusion. Visualization Generation A visual agent creates Python code using matplotlib to produce contextual data visuals. These are executed and displayed dynamically in the UI. Social Media Post Creation A social copywriter agent distills the article into a 150–200 word LinkedIn post with hashtags and a call to action. 🔗 LinkedIn Authentication & Auto-Posting Users authenticate via LinkedIn OAuth (handled within the app UI). Once authenticated, the final LinkedIn post is published directly to the user’s feed using the LinkedIn API. Export Options Users can download the full article as a PDF or copy it to clipboard. Visualizations are also saved as image files. 💻 Technologies Used Streamlit – Interactive UI and user controls CrewAI + LangChain – Multi-agent orchestration & LLM-powered task execution OpenAI GPT-4 – Core language model for topic generation, research, writing, and posting LinkedIn API (OAuth + UGC) – Social media publishing integration Matplotlib – Data visualization rendering Python – Backend logic, REST handling, and agent coordination 🚀 Outcome A fully automated tool that: