Business Case: https://sdw.solutions/case-studies/ai-agent-truefit
Unlike standard LLMs, TrueFit combines multiple capabilities:
🧠 Long-term memory: It stores and recalls user information such as values, interests, and resumes using TrustCall. 🏢 Company extraction: It analyzes company data, extracting mission, values, and job details. 🔎 Tavily-powered RAG: When values or mission are missing, it uses Tavily search to enrich context by pulling fresh web content. ✍️ Tailored generation: Merges profile + company context to craft or evaluate cover letters. 🔄 Feedback loop: It can revise and score a letter against the job description. 🧾 Dual LLM usage: While GPT-4o powers extraction and generation, the review process uses Claude Sonnet for scoring and actionable feedback. 🛠️ Built with LangGraph + LangSmith: The entire application is designed as a composable LangGraph, enabling traceable logic, live debugging, and rapid iteration. LangSmith integration offers observability and monitoring during development. This level of contextual personalization goes beyond typical LLM prompting. TrueFit acts as a dynamic agent with both memory and search to give you the best cover letter possible.
Source code: https://github.com/nikpalumbo/truefit
