Case study · Project 01 / 04
AI Research Agent
A planner breaks a question into angles, parallel researchers dig in, and a synthesis step writes it up. Fully local.
- Role
- Workflow design, build
- Stack
- n8n, Ollama, Agentic workflows, Prompt engineering
What is AI Research Agent?
The AI Research Agent is an agentic workflow built with n8n and Ollama that runs fully locally. A planner breaks a research question into angles, parallel researchers investigate each angle, and a synthesis step merges what they found into one written report. It was designed and built by Shashank Joshi.
What problem does it solve?
Research questions have to be broken down, chased and written up by hand, every time. That repeated, multi-step work is what an agentic workflow can take over. The agent runs the whole loop, from the first question to the finished write-up, without anyone steering each step.
How does it work?
A question enters the workflow as a trigger. A planner decides which angles to investigate, three researchers work on those angles in parallel, and a synthesis step merges their results into the final report. Every step runs on local Ollama models, orchestrated by n8n.
- 01QuestionA question starts the workflow as a trigger.
- 02PlannerPlans the angles worth investigating.
- 03ResearchersThree researchers work the angles in parallel.
- 04SynthesisMerges the findings and writes the report.
What is it built with?
It is built with n8n for orchestration and Ollama for local models, using agentic workflow design and prompt engineering to coordinate the planner, the researchers and the synthesis step.
What does it show?
Agentic AI and automation skill: multi-step orchestration across several cooperating model calls that runs without hand-holding, built from open tooling and running fully locally.