Startup Stream

From a small town in Gujarat to  to the forefront of Silicon Valley’s AI revolution, Nirman Dave’s journey reads like a blueprint for the kind of ambition that defies borders, playbooks, and probability. Long before AI was a boardroom buzzword, Nirman was a teenager teaching himself to code by watching YouTube videos on a borrowed laptop. He didn’t stop at learning. By the time he was 17, he had built over 200 apps—used by more than a million people worldwide—and even launched CircuiTricks, a business that turned his passion for electronics into a profitable startup.

“I was always fascinated by the idea that code could do something real in the world—solve a problem, help someone, grow personally & professionally,” he once shared. “I didn’t know what a startup was, I just knew I wanted to build.”

That spirit carried him to the U.S., where he studied Computer Science and Behavioral Economics at Hampshire College. There, he didn’t just attend classes—he built movements. HampHack, the first interdisciplinary hackathon in the country, was his brainchild, created in partnership with Google and Viacom. For Nirman, every project was a playground for systems thinking and user empathy.

In 2020, he co-founded Obviously AI, a bold response to a question many were too afraid to ask: What if anyone could build a machine learning model—without writing a single line of code? The platform went on to power 82,000+ prediction models, saving companies over 12.8 million hours of data science work. More than 3,000 businesses used it to make everyday decisions smarter—from churn prediction to lead scoring.

“We didn’t want to just democratize AI—we wanted to make it feel like magic,” Nirman said in an interview with MJF. “You upload a spreadsheet, and 30 seconds later, you get a prediction. That’s the kind of simplicity that changes behavior.”

Obviously AI earned Nirman a spot on Forbes 30 Under 30, and drew investment from some of the world’s sharpest funds—Sequoia Scouts, Facebook’s startup fund, B Capital, and TMV among them. But even as the no-code AI category boomed, Nirman’s mind was already racing toward the next frontier.

That next leap became Zams, his current company, and an emerging powerhouse in enterprise automation. If Obviously AI was about giving startups superpowers, Zams is about giving scale to the Fortune 500—without the drag of manual work. The name Zams is short for Zero Age Main Sequence, a term in astrophysics that marks the moment a star starts to shine.

“Zams is that ignition point,” Nirman explains. “Where AI stops being a pet project and starts being infrastructure.”

Today, Zams agents automate complex back-office tasks for some of the world’s largest companies—from RFQ parsing to quote generation—freeing up teams to focus on high-value work. One client added $10M in new revenue without hiring a single new person. The platform’s goal? To make deploying an enterprise AI agent feel as intuitive as building a website on Webflow.

And it’s working.

Behind the scenes, Zams is powering the kind of automation that makes operations not just efficient, but intelligent. And Nirman’s approach is refreshingly pragmatic. He doesn’t promise a utopian AI future—he ships real tools, solves real problems, and listens obsessively to the pain points of his customers.

“AI doesn’t need more theory,” he says. “It needs better design. And a hell of a lot more empathy.”

What sets Nirman apart isn’t just the scale of his ambition—it’s the clarity of his conviction. He’s a founder who’s equally at home in product reviews, investor pitches, and late-night build sessions with his team. He speaks often about the invisible work of company-building, the tension between vision and iteration, and the underestimated power of persistence.

As a fellow at INK and the Rajeev Circle, and a familiar face on AI panels and community events in SF, he continues to share that conviction freely. And yet, he stays grounded—always in builder mode, always two steps ahead.

For Nirman Dave, AI is not just a technology. It’s a tool of leverage. A creative force. A democratizer of opportunity. And most of all, it’s a blank canvas for the next generation of builders.

“The real win,” he says, “is when someone without a tech background builds an agent that changes how their company works. That’s when I know we’re doing our job.”

India’s quick-commerce revolution has transformed how people order groceries, cabs, and snacks — but urgent medicine delivery has lagged behind. Kaushal Shah, founder of pharmacy tech platform eVitalRx, saw an opportunity hiding in plain sight. In early 2025, he launched PillO — a platform promising 60-minute delivery of critical medicines, wherever you are — at home, at a railway station, even stuck in traffic.

Shah isn’t new to this world. Before PillO, he spent years inside India’s fragmented pharmacy ecosystem, helping digitize over 7,000 retail pharmacies through eVitalRx. That inside view — not just of tech, but how medicines physically move across Indian cities and small towns — gave him an unfair advantage when it came time to reimagine healthcare delivery for the on-demand era.

Unlike traditional e-pharmacies that rely on slow-moving warehouses, PillO taps into a dense hyperlocal network. The model feels refreshingly simple: partner with neighborhood pharmacies, build real-time logistics tech, and get life-saving medicines to consumers before a minor issue becomes a major emergency. Early adopters are noticing — like Haris, who shared on social media how PillO delivered urgent meds straight to his train coach at Bengaluru Railway Station within 36 minutes. Or Santosh, who managed to get critical medicines delivered mid-transit. These aren’t PR stunts — they’re genuine signals of unmet demand.

The timing couldn't be sharper. India’s ₹4 lakh crore pharmacy market is formalizing fast, while post-COVID behavior has shifted permanently. Blinkit and Zepto have trained Indian users to expect groceries in 10 minutes. Healthcare — especially urgent medicines — is the next logical leap. PillO’s bet is that "meds in minutes" will become a non-negotiable layer in how we live, move, and manage health.

But this isn’t just another fast-delivery story. Kaushal Shah’s advantage is upstream: the tech stack. For five years, his team built eVitalRx — a full-stack ERP used by pharmacists for 8–10 hours a day. It tracks real-time inventory, enables compliance workflows, and powers AI-based QA systems. Without this infrastructure, he believes no “partner pharmacy” model can scale with reliability. “Others built apps. We built plumbing,” he says.

As PillO scales to 15 cities over the next 18 months, the roadmap is expanding. From adding lab tests and partnering with insurance firms to enable cashless OPD experiences, the company is building toward a future where healthcare is as accessible — and reliable — as ordering a pizza.

I caught up with Kaushal Shah over breakfast at an eChai Social in Ahmedabad to learn what PillO is building. I asked a few questions. He answered them all, openly.

Medicines aren't pizzas. How do you balance the urgency of 60-minute delivery with the critical need for medical safety and regulatory compliance? Where is the real operational bottleneck today?

That is precisely the point. We almost never encounter a real scenario in life where we must have medicines in less than 15 minutes but at the same time we also don’t want to suffer long without it (in case of acute issues) or miss a dose (in case of chronic medicines). So we decided to pick a sweet spot of 60 minutes, where we also are not rushing our delivery partners that can jeopardize their lives. Also it gives enough time for pharmacists to run checks and balances on their end, to understand and verify prescriptions, cross-question patients if needed and make sure they are dispensing the right medicines. We deliver 90% of orders in 45 minutes anyways.

In terms of operational bottleneck, it's hiring and building our own delivery fleet in different cities of India. Other than that, we can scale quickly by adding a new city to our offering every month, which is faster than even the Q-Commerce giants.

Today, early users are sharing amazing stories. As you scale across cities and handle thousands of orders daily, how will you maintain the same quality and reliability of experience? What could break?

I think the answer is similar to the previous question: it's the delivery fleet that we need to keep building and keep delivering the best experience in the space where almost all of us have had horrible experience (at buying medicines when we need them) in the most important aspect of our lives: our health.

To answer how do we maintain the quality and reliability of experience from the pharmacy and dispensing side, we are adding new checks and balances every few weeks. I will share a couple of examples:

We have seen other startups attempt faster pharmacy deliveries before. Some succeeded briefly, many struggled. When you look at those previous attempts, what are your biggest takeaways or lessons?

Unfortunately, most of those startups think (and still do) that “partnering” with local pharmacies will work without putting in any work. What most of them do is make a low-touch app, onboard local pharmacies and start sharing the orders. There are three problems:

The app they need to make to engage with pharmacists has to be a deeper engagement with much broader information. That app is an ERP/Billing software that pharmacy uses for 8–10 hours a day and also has real-time stock information as well as billing mechanism for each of billing, accounting of those orders and payment reconciliation.

No other startup has invested their time and efforts to build this. This is what eVitalRx did for the last 4 years and it is due to this product and network, PillO is easier to build and is more likely to succeed than anything else out there.

Unless someone is willing to build this kind of software and better than eVitalRx for years, this is unlikely to work out.

The other option, which has become now the “conventional” option is to: “we will just open dark stores like Blinkit.” Well to them, I have one question only: “Then how will you beat Blinkit or Apollo for that matter?” I think it's a ridiculous and absurd approach — and if it was that easy, Blinkit and Zepto would be doing pharma Q-Commerce for the last 2 years now. But they haven’t!

There is a lot of quick-commerce buzz lately. Why do you believe PillO is not just another fast-delivery story? Why you and why now?

Agreed — and that’s why discussions like this are important to differentiate the nuances. PillO has been in the making for the last 5 years essentially. We launched PillO 3 months ago but foundations are being built for the last 5 years, before pretty much any of us knew the term “Quick Commerce”. Don’t believe me? Come to my house in Ahmedabad and right outside you will see a poster of eVital from 5 years ago promising the same thing that PillO does now.

Only thing we were missing is the very ERP system we have now. Meaning, we were just as naive as other medicine delivery apps are naive now.

Why me? Anyone can do this, they just have to commit to build the best ERP, POS and CRM software the pharma space has ever seen in India along with a centralized and standardized product catalogue — which one very senior manager from Udaan called “insanity.” Other than that, my experience of visiting, selling and talking to 1000+ pharmacists in India helps a lot.

Why now? I think one thing that has changed about consumers in India after Q-Commerce arrival is that people have started paying for convenience and we hope that if people are paying delivery convenience for pizza, they might do the same about their health. Also, I think small business owners — i.e., pharmacists and pharmacy owners — have also accepted the shifting business nature and have welcomed technology and collaboration.

What is the way forward for PillO? Beyond 60-minute deliveries, what is the bigger roadmap you are building toward?

First, we will be scaling to the top 15 cities of India in the next 18 months. Beyond that, we are also adding lab tests through our partnerships with other players and also collaborating with insurance providers to provide a cashless experience to patients for their OPD claims etc.

..

Several other players have recently entered the quick-delivery medicine space. PhonePe’s Pincode has started piloting 10-minute medicine deliveries in Bengaluru, Mumbai, and Pune by working with local pharmacies. Zepto has launched Zepto Pharmacy as a pilot across parts of Bengaluru, Mumbai, and Hyderabad. Flipkart is reportedly preparing to expand its ‘Minutes’ service to include medicines, also through local partnerships. These moves signal growing interest in adding healthcare products to the quick commerce mix.

The real test for PillO won’t be 60-minute deliveries — it’ll be scaling trust across cities, partners, and patients. If it works, Kaushal Shah’s legacy may not just be another healthtech play, but the foundation of India's true health-on-demand infrastructure. And this time, the infrastructure might arrive faster than the medicine.

In the fast-moving world of tech, few have traversed the founder-to-VC arc with as much depth, grace, and impact as Mala Ramakrishnan.

She’s not just funding innovation — she’s lived it, coded it, scaled it, and sold it. And today, she’s using that hard-earned wisdom to back the next wave of AI-first startups through Progressive Ventures, the seed-stage firm she founded with a simple thesis: “Back builders. Believe early.”

“I’ve sat on every side of the table — founder, operator, investor. I know what it’s like to pitch with hope and lead with grit,” Mala shares. “That perspective keeps me grounded in what really matters — clarity of vision, obsession with the problem, and founders who are all-in.”

..

Before stepping into the world of venture capital, Mala was a three-time founder with successful exits — proof that she doesn’t just analyze product-market fit from a distance; she’s fought for it in the trenches. Her resume includes leadership roles at WhatsApp (Meta), Cloudera, and Automation Anywhere, where she built and scaled enterprise platforms in machine learning and privacy.

At WhatsApp, she led Consumer Privacy. At Cloudera, she launched AI and ML platforms. At every stop, she paired technical depth with product instinct — and the rare ability to translate between engineers and execs.

“The hardest problems in tech aren’t technical — they’re human,” she says. “Alignment, storytelling, trust — those are the real levers.”

..

Through Founders Creative, the nonprofit she co-founded and now presides over, Mala supports over 10,000 underrepresented technologists — a testament to her belief that innovation must reflect the diversity of the world it hopes to serve.

“The talent is there. The ideas are strong. What’s often missing is access,” she says. “We’re changing that — by building community, not just capital.”

Her work sits at the intersection of deep tech, inclusive access, and raw conviction. It’s no surprise that founders describe her as “the VC who gets it.”

...

“She’s the kind of VC who doesn’t just ask the hard questions — she helps you answer them,” says one founder she backed pre-seed.

Another simply calls her, “the rare investor who’s walked the walk.”

..

In a startup world often filled with polished decks and performative pitches, Mala stands out as someone who cuts through the noise. She doesn’t buy into hype. She listens for substance. And she shows up with the kind of clarity that only comes from having been in the arena herself.

“You can’t fake founder energy. I back people who can’t not build.”

Whether she’s writing the first check or mentoring a founder on the brink of burnout, Mala brings the same principles to the table: honesty, empathy, and an unapologetic bias toward execution.

And for those lucky enough to share a room with her, one thing is certain — you’ll leave with more than advice. You’ll leave with perspective.
..

If you want to go deeper into Mala’s journey — from Stanford to scaling deep tech, from building enterprise products to backing the next wave of B2B AI startups — her recent conversation on the Verso Talk with Aytakin podcast is a must-watch. It’s a candid, thoughtful look at how she blends technical depth with founder empathy, and what drives her to back bold ideas early.

https://youtu.be/i1lD1qZVZbo?si=cTjnYkXhUmmJmkUc

On May 28th, I’ll be hosting Mala in San Jose at an eChai gathering we’re calling The AI Gold Rush: Behind The Bets. She’ll be joining us for a candid fireside chat, sharing what she’s seeing in early-stage B2B AI — the patterns, the GTM shifts, the blind spots most founders miss, and what real traction looks like in 2025. If you're building in AI, product, or anything that requires belief before proof, this is the kind of room where clarity lands and conversations linger.

Because some VC conversations don’t jsut belong in boardrooms.

They belong in rooms like this — where the stories are real, the rooms are warm, and belief doesn’t need a filter.

I recently rewatched the AlphaGo documentary, and even though it originally came out in 2016, it feels more relevant than ever in 2025.

At its core, AlphaGo is the story of how DeepMind’s AI took on the complex and ancient game of Go—and beat one of the world’s best players, Lee Sedol. But it's more than just a documentary about a machine winning a game. It’s a powerful exploration of human creativity, intuition, and what happens when we meet a new kind of intelligence.

One of the moments that gave me chills was when AlphaGo made a move that shocked even the experts—a move that no human would’ve ever played, yet it turned out to be brilliant. That moment felt like a turning point, not just in the game, but in the way we see AI: not just as a tool, but as something that can innovate in its own right.

What really hit me, though, was watching this now—nine years after AlphaGo’s big moment. In 2025, AI is everywhere. We now use it in our daily lives without even realizing it: from content creation and medicine to personal assistants and scientific research. Watching AlphaGo today feels like looking back at the first step of a long, fast-moving journey. It was a glimpse of the future—one that has now arrived.

But despite all the advancements, what makes this documentary so special is how human it is. You feel the pressure Lee Sedol is under. You see the awe in the eyes of developers watching their creation do the unexpected. You experience the quiet philosophical shift as people start to ask: What now?

If you’ve never seen AlphaGo, I highly recommend it—not just for the tech, but for the reflection it invites. And if you have seen it, watch it again. In light of everything AI has become since then, it hits differently.

This weekend, I sat down to watch Karpathy’s 3-hour deep dive into how large language models like ChatGPT are trained — and it turned out to be absolutely worth every minute. What surprised me most? It’s not just for AI experts.

Karpathy explains everything in a simple, clear way that anyone can follow — even if you're new to machine learning.
If you've ever wondered how AI actually works, this is the one video you shouldn't miss. It's insightful, beginner-friendly, and a perfect weekend watch for anyone curious about the future of technology."

From tokenization and attention mechanisms to model training and inference tricks — this isn’t just a talk. It’s the ultimate roadmap to understanding large language models.

Here’s how AI like ChatGPT is actually trained — in three essential stages:

1. Pre-training:

The model starts by learning from massive amounts of internet text — books, articles, code, Reddit posts — to build a general sense of how language works.

2. Supervised Fine-Tuning:

Next, it’s taught to behave like a helpful assistant by studying high-quality, human-curated conversations. This gives it purpose and structure.

3. Reinforcement Learning (RLHF):

Finally, the model is rewarded for good answers through trial and error. It improves by figuring out which token sequences work best — sometimes even inventing clever reasoning strategies.

But it’s not perfect! Karpathy also dives into the “sharp edges”:

Hallucinations (making stuff up)
Trouble with basic tasks like counting or spelling
Dependence on prompting techniques to stay accurate

>> Tools, better prompts, and smarter training are all part of making LLMs more useful — and Karpathy’s breakdown shows exactly how it's happening.