Soumava Dey

Building Autonomous Research Agents

Soumava 'Sam' Dey

15+ years in AI/data analytics at CMI Media/WPP & AIG; featured AI influencer

Create a CrewAI system that autonomously finds, reads, and summarizes research papers, extracting key insights automatic

Build an Academic Research Agent That Reads Papers for You

Structure agent roles and collaboration patterns to handle complex research tasks like paper analysis efficiently.

Design Multi-Agent Workflows for Literature Review Tasks

Implement cost optimization, quality validation, and monitoring for research agents that deliver reliable results consis

Deploy Production-Ready Research Agents Beyond Prototypes

Learn to build AI agents and workflows—developing judgment for what works, what doesn't, and how to adapt as capabilities evolve.

AI Research Expert | Johns Hopkins PhD

Google

PayPal

Bloomberg

Capital One

John Whalen

Schedules alternate by cohort to support multiple time zones. This cohort meets afternoons (ET).

April 2026

June 2026

You've been prompting ChatGPT and Claude. Maybe you've also used Listen Labs, or Marvin. But you've hit two ceilings: prompting stops when the prompt does, and SaaS is built for typical cases — it doesn't bend to your team's methodology or your industry's specifics.This course is what comes next.You'll build custom agents that read transcripts the way you would, every time. Assembly lines where one agent extracts themes, another flags contradictions, a third drafts synthesis. Synthetic experts and users informed by your real research data — not someone else's training set.The shift is from AI user to AI system designer. From researcher to research orchestrator.You'll build research systems with validation, traceability, and systematic evaluation. You'll learn when to trust autonomous execution — and when to keep humans in the loop. You'll deploy synthetic users with scientific rigor, not vibes. All using Claude Code or Codex: coding with words.No programming background required. Just willingness to think in systems.Course 2 of the Certified Agentic Research Specialist (CARS) credential. Take it alone, or pair with AI for Customer Research for the full designation.

Move Beyond Prompting. Build Agentic Research Systems.

Build AI Agents for Your Research Workflow

Create and Deploy Synthetic Users

Build Custom Agents and Multi-Agent Systems

Overcome the Technical Barrier

Orchestrate End-to-End Research Pipelines

Lead Advanced AI Research Initiatives

While not strictly required, Course 1 provides a strong foundation. You should be experienced with prompting AI and evaluating AI outputs.

Recommended: Completion of "AI for Customer Research"

Some agentic capabilities need Claude Code or Codex ($20/month). We'll guide you through setup.

Required: Claude Pro or ChatGPT Plus subscription

You'll work with an IDE, terminal, and automation platforms. We provide onboarding for non-technical researchers. New to me too! --John

Willingness to learn new technical interfaces

Alumni from "AI for Customer Research" ready to build autonomous agents, synthetic users, and workflows that work independently.<pre><code class="language-markdown"></code></pre>

Experienced researchers wanting to explore frontier capabilities—building agents, deploying synthetic users, and pushing beyond limits.

Product managers and designers ready to harness agentic AI for research—no coding background required, just curiosity to learn.

Become an Agentic AI Engineer by building real-world, production-ready AI agents that autonomously reason and execute complex tasks.

Cohort 2

Professor & AI Expert (540K+ Students)

Microsoft

General Motors

Samsung

HSBC

Dr. Ryan Ahmed

Professor, AI expert, & YouTuber who has taught 540,000+ learners worldwide.

Director of Engineering @Launchpad.AI

Nike

Levi's

Mr. Cooper

Launchpad

Kukesh Kodess

Director of Engineering with 7+ years building cutting-edge AI/ML systems

Cohort 1

Cohort 3

⭐"Join the instructor trusted by 540K+ students with 40K+ five-star reviews and 50M+ minutes of AI training watched worldwide."Agentic AI Engineering is 2026's most in-demand skill.Software is shifting from static systems to AI-native architectures, where autonomous agents reason, plan, and execute complex work.The problem: Most developers are stuck building API wrappers, and they lack the depth to manage state, memory, &amp; multi-agent coordination in real production systems.The Solution: This bootcamp is your bridge to elite Agentic AI engineering. We move beyond the basics to architect robust agent fleets in which specialized agents collaborate autonomously in production.What makes this course unique?<ul><li>Production-Grade Architecture: Engineer real systems using LangGraph, CrewAI, &amp; AutoGen.</li><li>Automate Agent Workflow Creation: Use Claude Code &amp; Codex to generate, refactor, and evolve agent workflows.</li><li>Enterprise Reliability: Design for guardrails, persistent memory, observability, &amp; human-in-the-loop safety. </li></ul><h3>Real-world Capstone Project</h3>Design &amp; deploy a production-grade, portfolio-ready agentic AI system for real-world use, demonstrating senior-level Agentic AI engineering skills.

Master Agentic AI Engineering & Architect Production-Grade Autonomous Systems

Architect, build & deploy production-grade agentic AI systems end-to-end

Orchestrate Autonomous Multi-Agent Teams

Master Tool Use & The Model Context Protocol (MCP)

Automate Workflows with AI Coding Assistants

Engineer Context-Aware Agents with Retrieval Augmented Generation (RAG)

Deploy, Monitor & Benchmark for Enterprise

Software Engineers &amp; Developers wanting to master state-of-the-art Agentic AI frameworks and build autonomous, production-grade AI agents.

Technical Founders &amp; CTOs who want to architect scalable, fault-tolerant autonomous systems that serve as your product's core engine.

Modern ML &amp; Data Science Practitioners who want to evolve from static models to dynamic Agentic AI workflows that reason, plan, &amp; execute.

How to build and run agentic AI research systems that generate structured product intel without manual work or headcount.

Principal Researcher, HXRlabs

Novartis

Johnson & Johnson

Pfizer

AliveCor

Myriad Genetics

Eric Shumake

20+ years in ethical product research and strategy- Principal at HXRlabs.com

You have been operating without enough research for too long. Not because you don't care about evidence. Because the system was never built to give it to you. No researcher on staff. No budget to hire one. No time to do it yourself. That changes here.Over three live sessions you will build a complete agentic product research operation from the ground up. Not a demo. Not a template. A real system configured to your product, your competitors, and your research questions that runs on a schedule and returns structured intel without you in the loop.Week one you build your first agent and learn what makes it actually work. Week two you expand into a full research stack covering competitive sweeps, user signals, and market trends. Week three you operationalize it, schedule it, and present your live system for direct feedback and coaching.By the end you will have something most product teams don't: a research operation that compounds in value the longer it runs. We use Claude Cowork to build it.This is not just a course about AI. It is a course about becoming the kind of product person who never has to make a call without evidence again.

Three Weeks to a Research Operation That Runs Itself.

Build Configured Research Agents

Expand Into a Full Research Stack

Operationalize and Automate the Loop 

Product managers accountable for roadmap decisions who need research they can generate themselves without a researcher or budget approval.

Designers and researchers embedded in product teams who spend too much time hunting for intel and need a reliable, automated system.

Founders and solo operators running product without research support who need a system that generates evidence while they focus on building.

Stop doing research manually. Build AI agents that do it for you.

20+ years in ethical healthcare design - Principal at HXRlabs.com

You already know what it feels like to ship a decision without enough evidence. You did the best you could, but somewhere in the back of your head you knew the research wasn't there. That's not a discipline problem. That's a resourcing problem. Agents can solve it.This workshop is three hours of live building against your real product context. You come in with a product, competitors, and a research question you have been putting off. You leave with a fully configured, tested research agent that runs on a schedule and returns structured intel you can actually use.The difference between this and every other AI workshop is that you do not just learn how agents work. You build one that works for your specific product, test it live with coaching, and walk out with something running before the session ends. We use Claude Cowork to make it happen.No homework after. No prompts to clean up later. No concepts to implement someday. One working agent, built and validated before you close your laptop.The bar for product decisions is rising. The ones who generate real research without a dedicated researcher will make better calls and outpace everyone still doing it by hand.

Product managers and owners who make roadmap decisions without dedicated research support and need evidence they can generate themselves.

Designers and researchers embedded in product teams who spend more time collecting intel than acting on it and need a faster system.

Founders and solo operators who wear every hat and need research that runs without them so they can stay focused on building.

Agentic AI

Coding with AI

AI Workflows

Claude Code

OpenClaw

Vibe Coding

AI Evals

AI Transformation

RAG & Search

AI for PMs

AI for Engineers

AI for Designers

AI for Marketers

AI for Founders

Product Sense

Product Discovery

User Research

Prototyping

Growth

Analytics

Tech Foundations

Strategy

Influence

Leadership

Career Growth

Machine Learning

LLM Ops

Context Eng

Security

System Design

Figma

Design Systems

Visual Design

Design Strategy

Automation

Content Marketing

Demand Gen

Go-to-Market

Product Marketing

Positioning

Social Media

Brand

B2B Marketing

SEO & AEO

AI for Leaders

AI Governance

Communication

Management

People Operations

Exec Presence

Storytelling

Goal-setting

Personal Brand

Fundraising

Everyone

Operators

Data Scientists

Business Analysts

User Researchers

Customer Success

Project Managers

HR Professionals

Sales People

Lawyers

Finance

Investors

Real Estate

Educators

Creators

Introduction to the Speaker and Topic

Speaker's Background and Expertise

Session Overview: Building an Autonomous Research Agent

Challenges of Enterprise AI Agent Demos

Understanding the Crew AI Framework

The Core Components of Crew AI: Agents, Tasks, and Crew

System Architecture and Mitigating Hallucination

Deep Dive into Agent Configurations

Code Walkthrough: Setting Up the Crew AI System

Code Walkthrough: Customizing Prompts and Tools

Code Walkthrough: Building the Streamlit UI

Live Demo: Researching AI Agent Platforms

Live Demo: Researching Transformer Architecture

Live Demo: Researching AlphaFold and Handling Firewalls

The Value of AI Agents in Academic Research

Q&A: Open Source and GitHub Repository

Q&A: Running the Agent and Future Web Version

Conclusion and Final Remarks

Broad professional audience — content relevant across many roles and functions.

Startup founders building new businesses, from MVP to scale.

Applying AI tools and techniques to professional workflows — assign alongside more specific AI topics.

Building autonomous AI agent systems that take actions, use tools, and orchestrate multi-step workflows.

Building Autonomous Research Agents

What you'll learn

Why this topic matters

You'll learn from

Soumava 'Sam' Dey