Cohort 1 is now open. · September 8 – October 1, 2026 · 4 weeks · Ages 18–21 · Reserve a seat →
Young Builders LabAges 18–21 only4-week live cohort · Sep 8 – Oct 1, 2026
At 19, anyone can prompt AI. Almost nobody can ship it and get users.
Four weeks, built for young builders. Week one is foundations — how these models really work, and the taste to tell good from generated. Then three builds and three launches: they build it Tuesday, they sell it Thursday.
If you're 18–21: you'll leave with three products live on the internet, launches that actually ran, and real people using them. Any stream. No coding background needed.
If you're the parent: they learn both halves — building with AI, and getting people to use what they built — plus the judgement to know what is worth building at all. The strongest builders get our backing to take it further. Parents, start here →
Build nights are terminal, Claude Code, databases, agents, and a real deploy. Sell nights are positioning, a landing page, a launch post, and reading the analytics on who actually turned up. Underneath all of it: knowledge, craft, judgement, and taste — the parts that do not come free with the tool.
“Learn to sell. Learn to build. If you can do both, you will be unstoppable.”
— Naval Ravikant
Almost every course teaches one half. Coding bootcamps teach building. Marketing courses teach selling. A 19-year-old who can only do one is still waiting for someone else to do the other. This cohort is built on the assumption that both are learnable in four weeks, because with AI, they now are.
Every Tuesday
They build.
Claude Code, the terminal, a database, an agent, a deploy. Something real exists by the end of the evening.
Every Thursday
They sell it.
Positioning, a landing page, a launch post, the channels, the analytics. It goes out in public and real people show up.
One week of foundations, then three builds and three launches. Both halves, every single week.
The Gap
Using AI is now table stakes. Building with it is not.
Every 19-year-old can get an essay out of a chatbot. That skill is already worth nothing — everyone has it. The gap that matters is between the ones who consume AI output and the small number who point AI at a real problem, ship a product, and get strangers to use it.
Using AI
Building with AI
Asks a chatbot for answers and homework
Points Claude at a problem and ships a product that solves it
Has nothing to show a recruiter except a CV
Sends a link. The work is live and anyone can open it
Stops at the browser tab — never opens a terminal
Works in the terminal, with a database and a deploy pipeline
Waits for a course, a degree, or permission to start
Has already shipped four things and knows the loop by heart
Builds a project nobody outside the class ever opens
Launches it, posts it, and watches strangers sign up
Arrives at college with the same AI skills as everyone else
Arrives able to build the thing the group project needs — and get it used
This is not a lecture series about AI. Every session ends with something that runs, deploys, or ships — and the last one ends with a product live in public and the first users signing up. Session one included: they will have code on the internet before the first evening is over.
For Parents
You are not buying a hobby class. You are buying a head start.
Most parents we speak to say the same thing: my child should be genuinely capable with AI before college, not just fluent at asking it questions. Capable means two things — they can build it, and they can get people to use it. We teach both, and here is how you will know it worked.
Proof, not attendance
Three links and a user count
At the end you get URLs, not a certificate photo — one launched product a week for the last three weeks. Plus the analytics showing the strangers who actually used them.
Build and market
Both halves, with Claude
Tuesday they build with Claude Code. Thursday they use Claude to position it, write the landing page, and run the launch. Naval Ravikant put it best: learn to sell, learn to build, and you are unstoppable. Most courses teach one half.
Taught by founders
Not a TA
The theCRUX cofounders teach this cohort themselves. They have built and sold companies, and they have trained more than a thousand working professionals on these tools.
The one thing we ask of you
Do not register on their behalf and surprise them. This is a build cohort with homework between sessions, and it only works if the builder wants to be there. Show them this page. If they get curious, book the seat.
Your Instructors
Founder-taught. One room of 25.
Bharanidharan Viswanathan
Instructor · Co-founder, theCRUX
3X founder. Built 91mobiles, SaaSworthy, and Intoglo across 20+ years in B2C, B2B, and cross-border logistics. Went deep on AI in 2021 — the models, the maths, and what separates a real product from a demo. Teaches with a builder's impatience for theory.
CEG, Anna University · MBA, ISB
Shameek Chakravarty
Instructor · Co-founder, theCRUX
4X founder. 20+ years building products across startups and big tech — Amazon, Yahoo, Sun Microsystems, Juniper Networks. Founded Farmizen and two AI ventures. Has trained 1,000+ professionals in AI, and builds agentic AI applications today.
CS, BITS Pilani · MBA, ISB
Ramkumar R.
Instructor · Co-founder, theCRUX
CEO coach and AI educator. 20+ years working with leaders on decisions, teams, and performance. Not a developer — learned Claude as a non-technical professional and built his own tools with it. Brings a coach's patience to a room of first-time builders.
BE, DCE · MBA, ISB
What We Actually Teach
The tool is free. The judgement isn't.
Claude Code is available to anyone with a laptop. So the course cannot be about the buttons — it has to be about the four things underneath, the ones that separate a builder from someone with a subscription. We teach them explicitly, in week one, and then keep testing them in every build after that.
Knowledge
How it actually works
What the model is really doing when it answers. Why context decides everything. Why it invents things, where it is genuinely strong, and which jobs never go to it unchecked. Understand the machine and the surprises stop.
Craft
How to direct it well
The distance between a prompt and a spec is the distance between a toy and a product. Briefing, structuring context, iterating, and describing a failure precisely enough that it actually gets fixed.
Judgement
Knowing what's wrong
Spotting the confidently incorrect answer. Knowing when to push and when to start over. Deciding what to cut. And the first filter of all — whether a thing is worth building before a single line is written.
Taste
Knowing what's good
Anyone can now generate a hundred versions. Only taste picks the right one. We do side-by-side teardowns of good work and generated filler until they can feel the difference without being told — in the product, the design, and the writing.
Here is the honest reason this matters. In two years the tools will be better than anything we teach this September, and every syllabus written today will be out of date. Knowledge, craft, judgement, and taste will not be. That is what we are actually trying to leave behind.
Programme at a Glance
Four weeks. Foundations, then three builds and three launches.
Two evenings every week, two hours each. Week one lays the foundations — how the models work, and the craft and taste to use them well. Weeks two to four each produce one complete product: Tuesday it gets built, Thursday it gets positioned, launched, and put in front of real people.
🧭
Week 1 · Foundations
Sep 8 & 10 — How this actually works. What the models really do, how to direct them, and how to tell good work from generated filler.
🚀
Week 2 · Build 1
Sep 15 & 17 — The Portfolio Site. Tuesday they build and deploy it. Thursday they position it and put it in front of real people.
🧠
Week 3 · Build 2
Sep 22 & 24 — An AI App People Use. Tuesday: database, agent, deploy. Thursday: landing page, launch post, first users.
🏁
Week 4 · Build 3
Sep 29 & Oct 1 — The Capstone. Tuesday they build their own idea. Thursday they launch it in public and demo the numbers.
Session by Session
Foundations first. Then build Tuesday, sell Thursday.
Week 1 · Session 1
Tuesday September 8
6 – 8pm IST
Foundations
What Claude Actually Is
How these models really work — prediction, context, and why that explains almost every behaviour they will ever see
Context is everything: what the model can see, what it cannot, and why a great answer follows a great setup
Why it makes things up, where it is genuinely strong, and the tasks to never hand it unchecked
Chat, Claude Code, agents, MCP — what each one is for, in plain language
Terminal and Claude Code installed, and one small thing deployed live before they log off
Week 1 · Session 2
Thursday September 10
6 – 8pm IST
Craft & Taste
The Part the Tool Doesn't Give You
Briefing as a craft: the difference between a prompt and a spec, and why the second one gets built
Judgement — reading output critically, spotting the confident wrong answer, knowing when to stop and start over
Taste — side by side teardowns of good and generated work, until they can feel the difference themselves
Debugging as a conversation: how to describe a failure so it actually gets fixed
What is worth building at all — the first filter every builder needs and almost nobody teaches
Week 2 · Build
Tuesday September 15
6 – 8pm IST
Build Night
Build the Portfolio Site
Design and build a personal site worth sending to a recruiter, not a template with their name pasted in
Git and GitHub: how work gets saved, versioned, and shared
Custom domain, mobile layout, and a real deploy pipeline
Debugging with Claude: reading the error instead of fearing it
Week 2 · Sell
Thursday September 17
6 – 8pm IST
Sell Night
Make It Worth Sending
Positioning, on themselves first — what they want to be known for, in one sentence
Write the site with Claude: headline, proof, and a reason for anyone to care
Analytics installed before they log off
Send it to ten real people and watch the first visitors land in the dashboard
Guaranteed outcome — every builder ends week two with a live site people have seen
Week 3 · Build
Tuesday September 22
6 – 8pm IST
Build Night
An App That Remembers, and Acts
What a database actually is, and why every real product has one
Build an app with sign-in, stored data, and pages that change per user
Give it real tools with MCP so it can act, not just answer
Connect an external API and handle what happens when it fails
Deploy it, break it on purpose, and fix it live
Week 3 · Sell
Thursday September 24
6 – 8pm IST
Sell Night
First Users, Not First Marks
Write a landing page that converts — headline, proof, and one clear ask
Build the launch kit with Claude: demo video, screenshots, and a post that does not read like an ad
Where to post it: Reddit, X, LinkedIn, WhatsApp groups, campus channels
Ship the launch and read the analytics together — who came, what they clicked, where they left
Find the capstone problem, and talk to three real people who have it
Week 4 · Build
Tuesday September 29
6 – 8pm IST
Build Night
Build the Thing They Chose
Scope the capstone down to something shippable in one week
Build it live, with founders moving between screens
Code review on the spot — what to fix, what to cut, what to ship
Homework is the rest of the build; the group chat stays open all week
Week 4 · Sell
Thursday October 1
6 – 8pm IST
Launch + Demo Day
Publish It. Get the First Users.
Go live in the session — the launch post goes out and the first strangers arrive
Read the analytics on screen: who came, what they did, what to change tomorrow
Demo the product and the numbers to the room, with founder feedback on both
A plan for the next three things they build and launch on their own
What They Leave With
Three products. Three launches. One repeatable loop.
Not notes. Not a certificate. Three links you can open, built and launched by them, hosted under their own accounts — plus the dashboards showing the people who used them.
Week 2
A portfolio site, seen by real people
Live on its own domain, positioned and written with Claude, good enough to sit at the top of an internship application — and already sent to the first ten people.
Week 3
An AI app, launched
Sign-in, a real database, and an agent wired to real tools through MCP. Then a landing page, a launch post, and the first users who were not asked to sign up.
Week 4
A capstone of their own
A problem they picked, validated with three real people who have it, built over the final week, and launched in public on the last evening — with the analytics on screen.
Forever
The loop, and the judgement to run it
Build it, position it, launch it, read the numbers, fix it — run three times until it is muscle memory. Plus the week-one foundations underneath: knowing how the tool works, and the taste to tell what is worth shipping.
Beyond the Cohort
The best builders don't stop at demo day.
Every cohort, some projects show real signal — users who came back, a problem worth solving, a builder who kept going after the last session. We back those builders. Not with a certificate. With our time, our network, and a route to taking it further.
Step 1
Founder mentoring
Selected builders keep working with the three of us after the cohort ends — on product direction, the next version, and what the early users are really telling them.
Step 2
Help to incubate it
Where a project deserves to become a real thing, we help take it forward: shaping the first version of the business, the early go-to-market, and the practical steps of actually starting up.
Step 3
Our network, opened
Introductions to founders and operators we know, and a platform to put the work in front of people who can hire, use, or fund it.
Said plainly, so nobody is misled
This is selective, and it is judged on the work — not on attendance, and not on who paid. It is not a guarantee of funding, and we will not pretend otherwise. What we do guarantee is that a genuinely good project will not be left to die after the last session. Three founders who have built and sold companies will be paying attention to what your builder makes.
Who It's For
Built for 18-to-21-year-olds who want to make things
A strong fit if they…
Are 18 to 21 — in college, between school and college, or in a first job
Use AI daily but have never built anything with it
Have ideas they keep describing to friends and never building
Want people to actually use what they make, not just submit it
Study anything at all — engineering, commerce, design, arts. No coding background assumed
Can give two evenings a week for four weeks — a build night and a sell night — plus homework between them
Not the right fit if they…
Are under 18 — this cohort is 18 and above, and the pace assumes it
Want a computer-science curriculum with algorithms and exams
Are being signed up by someone else and do not want to be there
Are working professionals — Claude for Life is the cohort for you
Next Cohort
Cohort 1 · September 8 – October 1
Eight live sessions across four weeks, 6 – 8 PM IST, Tuesdays and Thursdays · 25 seats · Ages 18–21 · Bring an idea, or find one in week three. Either way it launches on October 1.
Launch Price · Save ₹20,000
₹29,999
Regular: ₹49,999
International: $349 (regular $599)
All taxes included
Capped at 25 seats
Eight live 2-hour sessions across four weeks — 16 hours, taught by the founders
A foundations week on how the models work, plus craft, judgement and taste
Three complete products, one per week, each one launched in public
Founder-instructors in the room, at their screens
Everything hosted under their own accounts, theirs to keep
Group chat with the cohort and founders for all four weeks
A shot at founder backing after the cohort — mentoring, incubation help, and introductions
Pay in ₹ (card / UPI) or $ (international card) · Claude Pro subscription required separately
What they need before September 8
Everything is built on their own laptop and their own accounts, so the setup below is a prerequisite. We send a short setup checklist as soon as you register.
Claude Pro
$20 / month
The free tier will not carry a build cohort; the Max plan is not needed. In India (₹2,399/month with GST) Anthropic bills by card, not UPI — so a parent's international card is usually the practical route.
A Laptop
Mac or Windows
Their own machine, with permission to install software. A phone or tablet will not work — this cohort lives in the terminal and a code editor.
Two Evenings a Week
For four weeks
Two hours a session, twice a week, plus roughly two to three hours of building between them. The capstone week asks for more. Builders who skip the homework will not finish the capstone.
FAQ
Common questions
No. We assume zero coding background and start with the terminal from scratch. Claude Code writes the code; the skill we teach is directing it, reading what came back, and shipping the result. Builders from non-engineering streams do fine — several of our strongest professional participants have never studied computer science.
The opposite. AI is the centre of the method, not a topic in it. Week one is entirely about how these models actually work and how to direct them well. Every build after that is done with Claude Code, and one full session covers agents, tools, and MCP — how to make AI do work rather than answer questions. What transfers is the ability to build with AI and the judgement to use it well, whatever they go on to study.
Both, from week two onwards — it is the whole structure. Tuesday is build night; Thursday is sell night. Selling covers positioning, a landing page that converts, a launch kit made with Claude (demo video, screenshots, and a post that does not read like an ad), where to post it, and how to read the analytics afterwards. They do this three times in three weeks, so it becomes a habit rather than a lecture. Week one comes first and teaches the judgement underneath it.
Through a payment gateway. At 18 and above a builder can sign up for one in their own name, or do it alongside a parent. Which gateway depends on where they are — Razorpay or Instamojo in India, Stripe or a local equivalent in Singapore and elsewhere. All of them ask for identity documents and a bank account before they release money, so that step often needs an adult in the room. The cohort covers where payment belongs in a product and how to add a payment link or checkout to what they built. The account and the money stay theirs; theCRUX never holds either.
Some launches land and some do not — that is the honest lesson, and it is worth more than a guaranteed number. What every builder leaves with is the machinery: a live product, a landing page, a launch kit, the channels to post in, and analytics that tell them what went wrong. We read the results together in the last session and they leave knowing the next move.
This cohort moves at adult pace, uses accounts with adult terms of service, and expects self-directed homework between sessions. We have kept it 18 and above so we do not have to slow it down. Email hello@thecrux.ai if you would like to hear when we open a younger format.
After every cohort we look at which projects have real signal — users who came back, a problem worth solving, a builder who kept going. Those builders keep working with the three of us: mentoring on the next version, help shaping it into a real business if it deserves that, and introductions into our network. It is selective and judged on the work, not on attendance. It is not a promise of funding, and we say so on the page. It is a promise that a good project will not be dropped the day the cohort ends.
Three working links: a portfolio site on its own domain, an AI app with a database and real tools behind it, and a capstone they chose themselves. All of it is hosted under their accounts and stays theirs after the cohort ends. The capstone is demoed live on the final evening, and you are welcome to watch that session.
Every session is recorded and shared. But each session builds on the last, so a missed evening means catching up before the next one. Week one especially — the rest of the cohort assumes it. If they already know they will miss two or more of the eight, wait for the next cohort instead.
Because a builder who does not understand the tool is guessing, and it shows in everything they make. Week one covers how the models actually work, why they fail, how to brief them properly, and how to tell good work from generated filler. They still deploy something live on the first evening — the foundations are taught through building, not lectured. But the judgement comes first, because in two years the tools will have changed and the judgement will not have.
Not on day one. Weeks one to three are guided — everyone learns the same foundations and builds the same two products. Finding a problem worth solving is taught at the end of week three, and the capstone is scoped with founder feedback before the final week starts. An idea they already care about is a bonus, not a requirement.
Most seats are paid for by a parent. Register with the builder's name and email so the joining details reach them directly, and pay with any card or UPI. For a GST invoice, email hello@thecrux.ai after booking with the GST details.
Full refund within 24 hours of booking. No refunds after that — the cohort is capped at 25 and a late cancellation takes a seat from someone else. If the dates stop working, email hello@thecrux.ai and we will move the seat to a future cohort.
No. theCRUX is an independent training company. We are not affiliated with, endorsed by, or sponsored by Anthropic — we are practitioners who use these tools daily and teach from that experience.
Young Builders Lab · Ages 18–21 · September 8 – October 1, 2026
In four weeks they go from asking AI for answers to launching something people use.
Four weeks. Eight evenings. Three builds, three launches. 25 seats, founder-taught, ₹29,999 launch price ($349 international).
