What can I build with AI as a complete beginner?

Tonight, with zero coding knowledge: a working personal tool - a budget tracker matched to your actual salary, a CV analyser, a study planner - by describing the problem to an AI assistant and asking for a single-file web tool you open in a browser. That’s rung 1 of a four-rung ladder. Within weeks of learning: small apps you actually understand and can fix (rung 2). Within months of structured learning: full applications with users, accounts, and payments (rung 3), and then AI-powered products where AI is the engine, not just the builder (rung 4). The ladder is honest about the catch: each rung requires more understanding, and skipping rungs produces demos that break.

Can I really build an app with AI without knowing how to code?

Yes for a working prototype; no for a product - and knowing where the line sits saves months of frustration. AI will genuinely produce a functioning app from a plain-English description tonight. But the no-code ceiling arrives fast: when it breaks (it will), when you need it to handle real users’ data safely, when you want to change something the AI misunderstands - you hit the wall of not understanding what you have. That wall is exactly where learning starts paying: people who can read, debug, and direct the code get compounding returns from the same AI tools that leave pure prompters stuck. The tools got everyone to the starting line; fundamentals still decide who finishes.

What are good AI project ideas for a portfolio?

The rule that beats any list: build for a real user you can name, solving a problem you’ve verified exists. A booking system for an actual tuition centre beats a tutorial-clone e-commerce site every time, because interviews are conversations about decisions - and real projects force real decisions you can defend. Strong shapes for 2026 portfolios: an internal tool for a family business or employer (inventory, quotations, rostering), a niche community product (industry job board, club management), and one AI-powered piece (document Q&A for a profession you know, an agent that automates a workflow you’ve done manually). Two or three deep, deployed, used-by-someone projects outperform ten shallow ones.

How long until I can build something people will pay for?

Honest range: roughly 4-9 months of consistent structured work for most people starting from zero - the time to climb rungs 2 and 3 properly (fundamentals, full-stack capability, deployment) plus the unglamorous part nobody puts in the timeline: finding a problem someone will actually pay to solve, which is a validation skill, not a coding skill. Documented solo founders compress this by building for niches they already know deeply - the accountant building for accountants skips months of discovery. What doesn’t work is inverting the ladder: going from zero to "charging for a rung-4 AI product" on prompting alone produces products that collapse at the first real customer. Our ai-product-builder programme and the solo-founders breakdown on our blog cover the full path.

Where should I start building tonight?

Two free options, matched to your mood. If you want the instant win: open any capable AI assistant, describe one small personal annoyance precisely ("build me a single-file web tool that..."), and have a working thing before bed - rung 1 takes an hour and the point is to feel the leverage firsthand. If you want the win that compounds: start the Sigmaschool free trial - one signup, no card - which gives you the 6 Projects in 6 Days precourse on Sigmo (rung 2: build real things while actually understanding them), a live Buildroom session with an instructor, and the full programme e-booklet. Best answer: do both - rung 1 tonight for the spark, rung 2 this week for the fire.

All posts

Build your own · AI & your future · Getting started

What can I build with AI? More than you think - starting tonight.

This is the excited version of the AI question, and it deserves a better answer than a listicle of app ideas. Here’s a four-rung ladder of what a normal person - not a genius, not an engineer yet - can realistically build, with honest time-and-skill labels on every rung: a working tool tonight, projects you understand within weeks, full applications within months, and the genuinely new thing at the top - products only this era makes possible for one person.

Deric Yee Updated 25 August 2026 9 min read

The thing that actually changed

For the entire history of software, there was a brutal gate between “I have an idea” and “a thing exists”: years of learning or tens of thousands of ringgit for a dev team. That gate is what AI removed - not the value of skill (we’ll get to that, honestly) but the cost of starting. In 2026 a person with zero coding knowledge can describe a problem in plain English and hold a working tool the same evening. The ceiling didn’t move down to meet everyone; the floor moved up to meet everyone - and the ladder between floor and ceiling is climbable in months. The documented proof at the top: solo founders running six- and seven-figure software products alone, profiled in our most-read article. What follows is the ladder itself, with the honesty most versions of this article skip: what each rung really requires, and why skipping rungs is the classic failure.

The ladder: four rungs, honestly labelled

Tonight (0 skill)

A personal tool that actually works

A budget tracker matched to your exact salary and commitments. A CV-vs-job-description analyser. A meal planner around your dietary limits. A study-schedule generator for your professional exam. Describe the problem to an AI assistant, ask it to build a single-file web tool, open the file in a browser.

The honest label: Zero coding knowledge required - this rung exists because AI collapsed it. What you get is real but fragile: it works until it doesn’t, and when it breaks you can’t fix it. That frustration is not a failure - it’s the reason the next rung exists.

Weeks 1-4 (learning begins)

Small projects you understand and can fix

A habit tracker that saves data properly. A quiz app for something you’re studying. A landing page for a friend’s business, deployed to a real URL. An expense splitter your housemates actually use. Built WITH AI as tutor and pair - but with you understanding every part.

The honest label: This is where "using AI" becomes "learning with AI" - the distinction our whole pedagogy hangs on. Slower than rung 1, deliberately: understanding is the product. Our 6 Projects in 6 Days precourse lives exactly here, and it’s inside the free trial.

Months 2-4 (real capability)

Full applications with users, accounts, and data

A booking system for a real tutoring centre. A community job board for your industry in Malaysia. An inventory tool for a family business. A client portal for a freelancer. Databases, authentication, payments, deployment - the full stack of a real product.

The honest label: This rung takes structured months, not inspired weekends - it’s the core of what a serious programme teaches, and it’s the rung employers hire from. The portfolio pieces that get interviews come from here: real problems, real users, decisions you can defend.

Months 4+ (leverage)

AI-powered products - the things only this era allows

A document-answering assistant for a niche profession. An agent that drafts quotations from WhatsApp enquiries. A content pipeline for a small brand. A tool that reads receipts and does the bookkeeping. Products where AI is the engine, not just the builder.

The honest label: This is the genuinely new territory - products a solo person could not have built in 2020 at any budget. Documented solo founders run six- and seven-figure businesses from this rung. It requires everything below it: rung 4 built on rung 1 skills produces impressive demos that fail in production.

Why the ladder beats the shortcut

The tempting move is obvious: if rung 1 works tonight, why not prompt your way straight to rung 4? Because each rung up adds something prompting alone can’t supply - the ability to know when the AI is wrong, and fix it. A rung-1 tool that breaks is a shrug; a rung-4 product that breaks has paying customers and their data inside it. This is the entire vibe-coding trap, argued in full in vibe coding vs fundamentals: AI multiplies whatever understanding you bring, and multiplying zero is still zero. The good news cuts the other way with equal force - the ladder that used to take a degree now takes months. Rung 2 to rung 3 is precisely what a structured programme compresses: 400–600 focused hours to job-readyis the honest number we’ve measured across our graduates, several of whom walked in at rung 0 and now hold developer roles at RM 6,000–9,000/month - the AI-capable junior band in our hiring report.

So start both fires tonight. The spark: build your rung-1 tool right now - one precisely-described personal annoyance, one working thing before bed, just to feel what leverage feels like. The fire: the free trial - one signup, no card - puts you on rung 2 this week with the 6 Projects in 6 Days precourse on Sigmo, a live Buildroom session with an instructor, and the full programme e-booklet. And if what excites you is specifically the top of the ladder - building a product rather than getting hired - the AI-Native Product Builder programme is the ship-to-graduate path for exactly that ambition. The era removed the gate. The ladder is right there.

FAQ

What can I build with AI as a complete beginner?
Tonight, with zero coding knowledge: a working personal tool - a budget tracker matched to your actual salary, a CV analyser, a study planner - by describing the problem to an AI assistant and asking for a single-file web tool you open in a browser. That’s rung 1 of a four-rung ladder. Within weeks of learning: small apps you actually understand and can fix (rung 2). Within months of structured learning: full applications with users, accounts, and payments (rung 3), and then AI-powered products where AI is the engine, not just the builder (rung 4). The ladder is honest about the catch: each rung requires more understanding, and skipping rungs produces demos that break.
Can I really build an app with AI without knowing how to code?
Yes for a working prototype; no for a product - and knowing where the line sits saves months of frustration. AI will genuinely produce a functioning app from a plain-English description tonight. But the no-code ceiling arrives fast: when it breaks (it will), when you need it to handle real users’ data safely, when you want to change something the AI misunderstands - you hit the wall of not understanding what you have. That wall is exactly where learning starts paying: people who can read, debug, and direct the code get compounding returns from the same AI tools that leave pure prompters stuck. The tools got everyone to the starting line; fundamentals still decide who finishes.
What are good AI project ideas for a portfolio?
The rule that beats any list: build for a real user you can name, solving a problem you’ve verified exists. A booking system for an actual tuition centre beats a tutorial-clone e-commerce site every time, because interviews are conversations about decisions - and real projects force real decisions you can defend. Strong shapes for 2026 portfolios: an internal tool for a family business or employer (inventory, quotations, rostering), a niche community product (industry job board, club management), and one AI-powered piece (document Q&A for a profession you know, an agent that automates a workflow you’ve done manually). Two or three deep, deployed, used-by-someone projects outperform ten shallow ones.
How long until I can build something people will pay for?
Honest range: roughly 4-9 months of consistent structured work for most people starting from zero - the time to climb rungs 2 and 3 properly (fundamentals, full-stack capability, deployment) plus the unglamorous part nobody puts in the timeline: finding a problem someone will actually pay to solve, which is a validation skill, not a coding skill. Documented solo founders compress this by building for niches they already know deeply - the accountant building for accountants skips months of discovery. What doesn’t work is inverting the ladder: going from zero to "charging for a rung-4 AI product" on prompting alone produces products that collapse at the first real customer. Our ai-product-builder programme and the solo-founders breakdown on our blog cover the full path.
Where should I start building tonight?
Two free options, matched to your mood. If you want the instant win: open any capable AI assistant, describe one small personal annoyance precisely ("build me a single-file web tool that..."), and have a working thing before bed - rung 1 takes an hour and the point is to feel the leverage firsthand. If you want the win that compounds: start the Sigmaschool free trial - one signup, no card - which gives you the 6 Projects in 6 Days precourse on Sigmo (rung 2: build real things while actually understanding them), a live Buildroom session with an instructor, and the full programme e-booklet. Best answer: do both - rung 1 tonight for the spark, rung 2 this week for the fire.

Back to all posts The top of the ladder, documented →

Build something real this week.
The gate is gone. The ladder is free to start.

Rung 1 tonight on your own; rung 2 this week with the free trial - 6 real projects in 6 days on Sigmo, a live instructor session, and the full programme e-booklet. One signup, no card.

Start the free trial Build a product instead