AI Studio Course

AI Agents and Agentic Web

MAS.664 / MAS.665 / 15.376 / EC.731 / IDS.865

(prev: Foundations of AI Ventures, AI Venture Studio, AI for Impact)



Overview

Build the agentic future
This MIT course teaches students to build autonomous AI agents that plan, coordinate, and execute complex workflows across distributed web systems. Students learn agent architectures, inter-agent protocols, web automation frameworks, and methods for decentralized coordination, while mastering AI-native coding and multimodal integration.

Through hands-on projects, students build sophisticated agentic applications: from automated workflow orchestration to intelligent web scraping and dynamic API coordination. The course combines solid technical foundations with innovative product development opportunities in this emerging infrastructure layer. Students learn to design AI agents with global impact, validate ideas through AI-assisted research and interviews, and move working prototypes toward market-ready solutions.

The course features insights from distributed AI researchers, web platform architects, and technical founders. It culminates with a technical demo day showcasing next-generation agentic web systems.

Course pillars
1. Innovation (20%): Cultivating a deep understanding of where AI is going to be in 2-3 years or more. Multimodal reasoning, autonomous workflows, personalized agents, robotics, and more.
2. Human Centered design and Final Project (30%): Learn how to design innovative AI products that can make a truly global impact, validate ideas with AI-powered outreach and interviews, for compelling market-ready solutions.
3. Technical Building (50%): We will teach you about hands-on creation of agentic systems, LLM-based tools, and automated pipelines using the latest AI stacks and frameworks.

Follow these steps to register for the course

  • Step 0: Register for course : For MIT (Link), For Harvard (Link)
  • Step 1: Questionnaire (Link 1) before Wed, Sept 3rd, 5pm
  • Step 2: NEW Build a mini AI app (Link 2) by Sun, Sep 7th

You must finish Step 1 to attend the first class. You will be expected to finish Step 2 by the second class.

Submission of two pre-homeworks does not guarantee acceptance, but you can send a request to consider your application in a form we will provide you in the class.

  • Pre-Course Info Session & Social: (Link, Link)

Class Schedule: Thursdays, 10 AM–12 PM, Room E14-633, MIT Media Lab (starting Sept 4)

Tentative Schedule

Course Instructors

Ramesh Raskar

Ramesh Raskar

MIT Media Lab
Serge Vasylechko

Serge Vasylechko

Harvard Medical School AI

Course TA

Leaders

Nikolay Vyahhi

Nikolay Vyahhi

Hyperskill and SundAI
Pradyumna Chari

Pradyumna Chari

MIT
Maria Gorskikh

Maria Gorskikh

Project Nanda
Frido Haugg

Frido Haugg

Harvard and SundAI
Konstantina Yaneva

Konstantina Yaneva

Harvard and SundAI
Honey Pamnani

Honey Pamnani

Lila.ai and MIT Alumni
Santanu Bhattacharya

Santanu Bhattacharya

MIT Media Lab
Elizabeth Beth Porter

Elizabeth Beth Porter

MIT Media Lab
John Werner

John Werner

MIT
David Blundin

David Blundin

Link Ventures

Mentors and Former Speakers

Alex 'Sandy' Pentland

Alex 'Sandy' Pentland

MIT Media Lab
Mark Weber

Mark Weber

MIT Media Lab and Tectonic Ventures
Nadia Shalaby

Nadia Shalaby

MIT and Pakira
Hossein Rahnama

Hossein Rahnama

MIT Media Lab/Flybits
John Harthorne

John Harthorne

Two Lanterns Venture Capital
Michael Hayes

Michael Hayes

Practical AI
Gargi Dasgupta

Gargi Dasgupta

Palo Alto Networks
Chris Pease

Chris Pease

Project NANDA
Taleb Alashkar

Taleb Alashkar

AlgoFace
Christy Fernandez-Cull

Christy Fernandez-Cull

Davinci Wearables
Mark Weber

Ana Pantelic

MIT D-Lab
Roman Cepeda

Roman Cepeda

Link Ventures and Exponential Management, LLC
Boaz Fachler

Boaz Fachler

Link Ventures
Narendra Kale

Narendra Kale

Kale Logistics
Jana Mike

Jana Mike

UCSF and BrainMind
Rajeev Surati

Rajeev Surati

Scalable Display Technologies and Skyline Nav
Mark Castleman

Mark Castleman

Intel Tiber Cloud
Brendan Mcsheffrey

Brendan Mcsheffrey

DataRight.org and The Kendall Project
Artem Lukoianov

Artem Lukoianov

CSAIL and SundAI
Pulkil Agrawal

Pulkit Agrawal

MIT EECS
Sertac Karaman

Sertac Karaman

MIT AeroAstro, Director of LIDS

Background

Almost 25% of students from this course transform their projects into innovative products. The program, which evolved from Sandy Pentland's "Development Ventures," has established a strong track record of launching successful student projects through its Demo Day showcase.

Our previous class links can be found here:

All other previous AI for Impact/Global Ventures class links can be found here:



Course Outline

Tentative Schedule

Class Goals:

  • Guide students to identify, evaluate, and build high-impact projects with AI. Focus on deep technology applications that are best suited to students with top engineering and design skills at MIT.
  • Advance "Big Ideas" into MVP-seed-stage for commercial consideration.
  • Teach the technical fundamentals of the modern AI stack (rapid prototyping, deployment, and model management).
  • Discuss practical trade-offs in AI product development (model accuracy, efficiency, and latency).

Timeline:

  • Meet and Greet held at MIT/Harvard
  • First day of class: Thursday, September 4, 2025
  • Final Class Presentations and "Demo Day" awards: Thursday, December 4, 2025

Class Structure:

  • Studio format, with team projects with 2-3 students per team
  • Guest speakers share leading-edge thinking
  • NB: You can use this course for Entrepreneurship Minor credits. Fill up this form under 'E&I Context' and meet E&I minor advisor, Reza Rahaman (rezar@mit.edu) for sign off.
  • Minimum Viable Product (MVP) by the end of the term
  • Demo Day

Support for Students:

  • Weekly class mentor meetings on Thursday after class / guest lectures / work sessions
  • Class visits to local start-ups and technology companies
  • Data and AI scientists from startups for experience
  • Sessions with startup coaches