Jan 16
Loading video player...
In this AI Flash session, Vanderbilt Data Science explores Anthropic’s newest Claude model: Sonnet 4.5-a major step forward in automated research assistance. Beyond generating text, Sonnet 4.5 plans, searches, executes code, wrangles data, and iterates, delivering sophisticated multi-step research and complex problem-solving. This workshop explores: How Sonnet 4.5 conducts multi-stage research: planning → tool use (web, code, files, APIs) → synthesis → verification. The distinctions between classic generative chat and agentic systems—and why reliable tool use and iterative planning matter. Evidence-first workflows: citations, source provenance, and methods for reproducible literature reviews and data analyses. Chaining tools for real tasks: web search + scraping, table extraction, coding and visualization, and structured note-taking. Safety, guardrails, and evaluation strategies for long-horizon reasoning (timeouts, retries, self-checks, and unit tests). Live examples include automated literature review pipelines, creating a lightweight RAG from fresh web sources, extracting + charting data from PDFs/CSVs, and building an end-to-end “research agent” that proposes hypotheses, runs analyses, and writes a concise brief. Whether you’re streamlining research workflows or tackling complex analytical problems, Sonnet 4.5’s long-context reasoning and trustworthy tool orchestration open new frontiers for applied AI in academia and industry. 📍 Learn More About AI Flash: vanderbilt.edu/datascience/ai-flash/ 📅 Subscribe for Future Sessions: bit.ly/vanderbilt-ai-flash