AutoResearcher
Seven-agent workflow that turns a prompt into a cited research report
Capstone project for Google’s 5-Day AI Agents course: a hierarchical, seven-agent system built with the Vertex AI Agent Development Kit and Gemini. It researches the web with Tavily and compiles a citation-rich report to PDF with Pandoc and LaTeX.
At a glance
- Status
- Course capstone
- Type of work
- Course capstone
- Context
- Google 5-Day AI Agents course
- Categories
- Applied AI & Agents
- Source code
- Public repository
Technology
Problem
Producing a sourced research report from a single question takes many manual steps.
Who it’s for
Course project; applicable to anyone who needs a first-draft literature or market scan.
What I built
- A seven-agent hierarchy built with the Vertex AI Agent Development Kit.
- Gemini 2.5 Pro and Flash for planning and writing; Tavily for web research.
- Pandoc and LaTeX to compile a citation-rich PDF.
Limitations
- Course capstone, not client or production work.