Too busy to listen? Here's the short version — Aarav Gupta turned 3-4 hours of nightly paper grading into EvaluMate, an AI grader. His biggest lesson? Stop assuming what customers want. Pick up the phone and ask.

What does building with AI actually look like beyond the hype? Aarav shares how he worked on AI solutions for the Bhutan government, explored agentic AI systems, and later co-founded Evalumate, an AI-powered education startup. In this episode, we explore problem validation, AI product development, and what it takes to build products from zero to one.

Aarav didn't plan to become a founder. He just got frustrated.

While running Titan Tuition, a small tutoring business, he spent hours every night correcting physics, chemistry, and math papers. The work was mechanical — matching answers to mark schemes. No creativity required. Just repetition.

When ChatGPT 4.0 dropped, he saw the obvious solution: automate it. He called his best coder. Brought on a tutoring teacher as an early investor. Evalumate was born.

In this episode of Lorong Product, we sat down with Aarav to unpack his journey from side gig to funded venture, his internship building AI for the Bhutan government, and why the metric that matters most is how fast you can fail.

Here's what stood out.

1. Problem First, Solution Second

Evalumate isn't for everyone. It's specifically built for IGCSCE and A-level curricula (CAIE boards).

Why that niche? Not because it's the biggest market, but because:

• Higher willingness to pay from schools and tuition centers

• Zero competitors already serving it

• Aarav understood the customer (he was one)

This is product thinking at its core. Pick a niche where customers are screaming for a solution, not where the market is biggest.

2. Friction Is Your Enemy

When Aarav demoed Evalumate, one feature stood out: WhatsApp integration. 

Teachers already use WhatsApp to send assignments. They don't want new logins, accounts, uploads. So instead of building a platform, he built a bot. 

Teachers forward assignments to WhatsApp → bot grades them → done.

He validated this by calling five teachers and asking directly. They said yes.

"Your initial hurdle to actually start using it becomes zero. That becomes the easiest onboarding."

3. The Mistake Every Builder Makes: Overbuilding

Aarav's biggest realization came when his co-founder pushed back on him wanting to optimize a feature.

"Don't tell me to change this feature until you know you have to change it from the consumers themselves."

Engineers default to optimization. Spending a month shaving 3 seconds off a response time. Users don't notice. Users don't care.

The fix: Talk to customers first. Then build.

"Ship, get feedback, iterate. Fail fast and move forward."

4. The NanoBI Internship: Building AI for the Bhutan Government

Aarav interned at NanoBI, building agentic AI solutions for government and financial services clients. His project: help Bhutan's government process documents faster.

The challenge? No AI model exists for Dzongkha (Bhutan's language).

He solved it using RAG (Retrieval-Augmented Generation) — a technique that grounds AI answers in real documents instead of letting it hallucinate. Here's his simple explanation:

"It's as simple as using a document to get certain information and putting an LLM on top to give a response. You're looking at lesser hallucinations, more based on that document itself."

By the end, he'd built a working frontend and backend framework the team could ship.

5. The Internship That Started With a Cold Email

Aarav applied to NanoBI through LinkedIn like normal. But he also did something bolder: he found the CEO and CFO's emails on the company website and sent them cold emails.

The CEO didn't respond. Someone from his team did. They became his manager.

The interview? Not technical. Just: "Can you learn?"

"They didn't expect anything from me in terms of technical expertise. They just wanted to know if I can learn or not."

6. Why Learning to Learn Is the Skill That Actually Matters

When asked what makes someone excel in AI roles, Aarav quoted Vinod Khosla (founder of Khosla Ventures):

"The skill you need to really learn is the skill to learn."

In 2024, being a generalist who can pick up new tools matters more than being a specialist. PMs code. Engineers design. Designers ship.

7. Product Court: Defending TikTok

In a fun segment, Aarav had 60 seconds to defend TikTok against charges of algorithmic manipulation and creator burnout.

His defense was pure product: the algorithm works because it listens to what users actually want, not what we assume they want.

"If a consumer likes the content and shares it, the best way to determine if a consumer is for it or against it is through the algorithm feed."

Verdict: TikTok found not guilty.

8. The Panic Mindset: Your Secret Weapon

Aarav's final advice isn't comfortable, but it's real.

Every founder should operate with urgency — what he calls the "panic mindset." Not anxiety. Urgency.

"If your product is actually good enough that it can make money, there will be somebody building it who's probably working harder than you. That needs to always ring in your brain."

In fact, if you're in a market with zero competitors, that's a red flag. Nobody wanting to buy is a bad sign.

8. The Panic Mindset: Your Secret Weapon

Aarav's final advice isn't comfortable, but it's real.

Every founder should operate with urgency — what he calls the "panic mindset." Not anxiety. Urgency.

"If your product is actually good enough that it can make money, there will be somebody building it who's probably working harder than you. That needs to always ring in your brain."

In fact, if you're in a market with zero competitors, that's a red flag. Nobody wanting to buy is a bad sign.

9. It's Never Been Easier to Start

Aarav's closing message: you don't need much to become a founder anymore.

The playbook:

1. Talk to 15 friends about what frustrates them

2. Find a pattern

3. Use Claude to build an MVP 

4. Make cold calls

5. Ship, get feedback, iterate

That's it. No massive capital. No need to be a technical wizard. Just customer obsession.

Takeaways

Aarav's story is a master class in product thinking without the jargon. He's a student, but he thinks like a founder who's been through three companies. He's building an AI product, but his obsession is entirely with customers, not technology.

The throughline? Ship fast. Talk to customers. Don't assume. Listen to feedback. Panic in the right way — with urgency, not anxiety.

His advice would resonate with anyone — founder or PM or designer — trying to figure out how to build things people actually want.

About the Author