AI Research Agents

Introduction

Research agents answer hard questions by searching the web (or your docs), reading multiple sources, synthesising, and citing. Perplexity is the canonical example.

Beginner analogy: Like a tireless analyst who reads 50 articles in 30 seconds and writes you a citation-rich brief.

Understanding the topic

Core concepts to understand:

• Plan sub-queries from the user question.
• Search (web, docs) in parallel.
• Read & summarise each source.
• Synthesize with citations.
• Reflect & deepen if confidence is low.

Syntax reference

Visual workflow / architecture:

Question
   │
   ▼
[Decompose into sub-queries]
   │
   ├──► [Search 1] ─► Read ─► Summary 1
   ├──► [Search 2] ─► Read ─► Summary 2
   └──► [Search 3] ─► Read ─► Summary 3
                              │
                              ▼
                         Synthesize + cite

Real-world use

Perplexity, You.com Research, OpenAI Deep Research, Google Gemini Deep Research.

Best practices

• Always cite sources.
• Cap parallelism to control cost.
• Use fast models for first pass, premium for synthesis.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. How does Perplexity work?
• Q2. What's the role of decomposition?
• Q3. Why cite sources?