For product practitioners and founders building a first product. Learn as faculty build an example AI product in five live online sessions. Use the materials and homework to build your own between sessions. Prework covers setup and basic tool use. No previous product launch required.
One part of your product that lets someone finish a task. Test it with evals, gather user feedback and identify what still needs work.
No previous product launch required
This cohort
Everything you need to know before you join.
Online
Cohort · November 2026
24 seats remaining
Location
Live online, delivered in English, available worldwide.
8:00 PM – 10:00 PM GMT+8Your local time · detecting…. Sessions run 8:00 PM – 10:00 PM GMT plus 8.
Dates
Build: 16, 18, 20, 23 & 25 Nov
Defense: 2 Dec
8:00–10:00pm SGT
Commitment
Five 2-hour build sessions, plus Defense Day
Price
SGD 499
Per seat.
Introductory price
Powered by
Includes US$100 Codex + US$50 API credits.
Ask for help in the community group throughout the course, with remote sessions when you need help getting past a problem.
Seats are confirmed in payment order.
Who this is for
For product practitioners and founders building a first product.
Bring a problem you want to solve. You do not need to have launched a product before. Prework helps you set up your accounts, tools and hosting, and practise using the tools before the live sessions.
You will fit if you…
Want to turn an idea or user problem into a small AI product
Want help with product decisions, design and building
Are willing to check AI-generated work, test it and make changes
Can make time for prework and between-session building
What to expect
Prework teaches basic tool use. You will still need to check and understand the work AI produces.
You will build with existing AI models. Training a model from scratch is outside the course.
You will learn design basics and how to make product decisions. The course does not teach a specific design tool.
Connecting company systems, using employer data and allowing unchecked changes to external systems are outside the course.
Product approach
From a user problem to a useful AI product
Learn to use AI tools to build, and decide how AI should work inside your product. An agent is one option. Choose the approach that best fits the problem.
Why AI?
Look at how people solve the problem today. Decide whether AI would improve the result.
What should AI do?
Decide what AI should read, create or act on, and where a person needs to check its work.
How does this help the user?
Name the task you are helping with and how you will check whether the product makes it easier.
What do you need to test?
Check that people have the problem, can use your product and get the result they need before adding more features.
What happens when it fails?
Use evals and traces to find mistakes. Decide how the product recovers and limit what it can access or change.
Curriculum and delivery
What you will learn.
Faculty build an example product in class. You help shape it through questions and product decisions, then use the same process to build your own between sessions.
How it runs
Before Session 1: set up your accounts, AI tools and hosting, and practise using the tools.
Sessions 1–5: watch faculty build an example AI product. Ask questions, discuss choices and help decide how the product should work.
At the end of each session: get the materials and homework for that part of the process.
Between sessions: use the materials and homework to build your own product and prepare for the next session.
Before Defense Day: you have one week after Session 5 to catch up, finish your work and prepare to explain what you built.
01
Prework
Get ready to build
Set up your accounts, AI tools and hosting. Practise using the tools, choose a user problem and look at how people handle it today.
Accounts and tools
Deployment setup
Tool familiarity
User and problem
AI foundations
02
Problem, value and success criteria
Decide what is worth building
Decide who the product is for and what it should help them do. Compare AI with other options, define what a good result looks like and list the assumptions you need to test.
User value
Why AI
Hypotheses and validation
Success criteria
03
Design and build a slice
Build one task from start to finish
Design and build the steps someone takes to finish a task. Include what they see while waiting and when something goes wrong. Add tools, memory and approval controls where needed.
Product design
Coding agents and SDK
Tools and state
Human controls
04
Deploy, trace and evaluate
Deploy your project and run evals
Put your project online and test how it behaves. Run evals for everyday use, unusual inputs and failures. Read traces to see the steps the AI took, and check cost, response time and permissions.
Deployment
Traces
Evals
Cost and response time
Permissions
05
Improve with evals and user feedback
Improve, then evaluate again
Use evals to check AI behavior. Ask people to try the product to find out whether it helps them and is easy to use. Make a change, rerun the same tests and note what still needs fixing.
User feedback
Rerunning evals
Failure and recovery
Making improvements
06
Defense Day
Show what you built and explain why
Demonstrate your project, explain your product and technical decisions, and share your eval results. Describe what still needs work before a small group of people could try it.
Your own project, test results and a process you can use again.
01
A working slice
One part of your product that lets someone complete a task from start to finish. This is your working slice.
A task someone can complete
Controls for reviewing or stopping AI actions
A way to recover when something fails
02
Evals and user feedback
Evals are tests of how your AI product behaves. Run the same tests after changes, and use feedback from people to check whether the product is useful and easy to use.
Tests for everyday use, unusual inputs and failures
Expected results compared with actual results
The same tests rerun after a change
User feedback and problems still to fix
03
A process you can use again
Use the course materials to plan, design, build and test your next project.
Choose a problem and check your assumptions
Reuse what you learned about design and code
Decide what to fix before trying a limited pilot
Defense Day
Show your product. Explain your decisions.
01
User problem
Who is this for, what problem does it solve, and why use AI? What have you learned from users?
02
Usability
Can someone finish the task and understand what to do if something goes wrong?
03
AI behavior and controls
What can the AI do, which tools can it use, and when must a person review or stop it?
04
How it works
How do the parts fit together, what data is saved, and how does the product recover from failures?
05
Cost and access
What data and tools can the product access? What does it cost to run, and how do you limit spending?
06
Evals and next steps
What did your evals and user feedback show? What needs fixing before a small group of people tries the product?
Your facilitators
Meet your faculty.
Learn from people with experience in product, design and engineering. They also review your work on Defense Day.
Deepika Murthy
Lead instructor · Product judgment & AI product building
Nineteen years across engineering, growth, marketplaces, platforms, and product. At Gojek, led product strategy for pricing, matching, batching, allocation, and fulfillment systems across Transport, Food, and Logistics, while driving GenAI adoption across the PM organisation. Earlier at Rakuten Viki, moved from video engineering into product and growth, launching and scaling a global subscription product to 200,000+ paid subscribers. Co-founder of Women in Product Singapore.
Choosing a problem, deciding what to build first, and explaining your product decisions using what you learned from tests and users.
Reviews on Defense Day
Scope
Judgment
Trade-offs
Commercial logic
Rajat Goyal
Applied AI and Agentic products
A builder across fintech, edtech, consumer apps, ride-hailing, and applied AI, with a decade of building and scaling production systems and the teams behind them. Most recently leading learning systems and applied AI at Aampe, scaling millions of per-user AI agents that personalize 1-to-1 and building agentic martech products. Earlier headed transport engineering at Gojek (millions of daily bookings across Singapore and Indonesia), and built financial and education platforms at Arcesium (D. E. Shaw Group) and Vedantu. Joins the Defense Day panel.
How the system works, how to monitor it, how to manage cost and security, and how to check whether an agent is ready for people to use.
Reviews on Defense Day
Production-readiness
Architecture
Cost & security
Observability & drift
Keith Oh
Product design
Product and design leader with experience in consulting and in-house teams. Most recently led Product Design and Core Marketplace product at Carousell. Earlier roles at IDEO, Rakuten Viki, and DSTA covered product strategy, design and research.
Honorary Secretary of Design Business Chamber Singapore, where he contributes to discussions on business, design and AI. His background spans Electrical & Computer Engineering, Psychology, and Human-Computer Interaction.
Designing screens and interactions, reviewing AI-generated designs, and checking whether people can use what you build.
Reviews on Defense Day
Taste
Craft
User clarity
Interaction logic
Design trade-offs
Ask questions about product decisions, design and engineering as you build.
Support & community
Get help while you build.
Ask the team and your peers for help between sessions.
Help in the community group
Post questions and discuss your work with the team during the week.
Optional live Q&A
Join optional live sessions when you need more help than a chat reply.
Stay in touch after the course
Keep learning with the people you meet in the cohort.
FAQ
Questions, answered plainly.
The answers here apply to this exact cohort.
I have built products before. Will this be too basic?
The course covers choosing a problem, designing a product, building with AI and testing the result. If you already do all of that confidently, you may need a more advanced course. Previous product-building experience is not required.
How much technical experience do I need?
You do not need to have launched a product before. Prework covers accounts, AI tools and deployment setup. We explain engineering concepts as we go. You will need to read, test and understand the work AI produces.
Do I need a product idea before Day 1?
It helps, but it is not required. Prework helps you find a user problem and understand how people handle it today. You will choose a small part of the problem to build for.
What do I actually own, and can I use it commercially?
The product and design work you create are yours to keep and use in your own work.
Do I need to have launched a product?
No. The course is for product practitioners and founders, including people building their first product. Set up your accounts, tools and hosting, and practise using the tools in prework.
What will I build?
One part of your own product that lets someone finish a task from start to finish. We call this a working slice. Faculty show the process on a shared example, and you use the materials and homework to build your own between sessions.
Will it be ready for a pilot?
Possibly. A pilot means trying the product with a small group of people. First, check your eval results, decide who can use it and what data it can access, and plan how to monitor it, help users and stop it if something goes wrong. Some projects will need more work.
Does every product need an agent?
No. You will learn when an agent helps, when a simpler approach is enough, and which actions need a person's approval.
Can I join the online AI agent course from outside Singapore?
Yes. Online cohorts are taught live in English and are open worldwide. Check the dates and the session times in your timezone before enrolling.
Is this an LLM course or an AI agent building course?
You will build a product using existing AI models and learn when to use an agent. The course covers product decisions, design, tools, memory, evals and controls for reviewing AI actions. It does not cover training a model from scratch.
Is this an AI certification course?
No. You will build a project and explain how it works. The course does not award an accredited certification or professional qualification.
Does the course cover AI transformation?
The course teaches you to build and test your own AI product. It does not cover changing how a whole organisation works or providing transformation consulting.
Is November enrollment open?
Yes. Enroll for SGD 499. Your seat is confirmed when payment is complete.
I already paid the S$10 priority deposit. How do I use it?
Enroll using the same email address you used for your November deposit. Your paid S$10 deposit is deducted automatically after any discount code, leaving SGD 489 without a discount. If the cohort is deferred, the deposit remains refundable.
Spread the word
Know someone who wants to build an AI product?
Cohort · November 2026 · Build: 16, 18, 20, 23 & 25 Nov · Defense: 2 Dec · Live online in English · available worldwide. Share the course with them.
