It’s Tuesday, and today we’re discussing Nūl, a climatech company out of Singapore founded by Malini Kannan and Raghav M.S. under the umbrella of venture builder Wavemaker Impact. The company launched with $500,000 in pre-seed funding.

Nūl is an AI solution that helps fashion brands reduce waste—mainly through two mechanisms: cutting overproduction and improving inventory management.

Both functions are executed by a series of agentic AIs, each with a specific role—forecasting demand, evaluating stock imbalances, optimizing store transfers, etc.

These agents make decisions based on two data streams: internal and external. Internal data comes from ERP, POS, and other systems—covering sales, production, stock levels, sell-through, and so on. External data includes weather, search trends, footfall patterns, marketing campaign performance, etc.

By combining and analyzing these large data sets in real time, Nūl helps at three levels:

Better optimization → less waste → higher profitability.

The last few weeks I found myself frequently writing about data and AI-related businesses—not intentionally, that's just how the chips fell.

But here's a realization I had: not all optimization is created equal. AI-driven software products can approach optimization differently, based largely on the nature and stability of the data they use. Here’s a way to clarify this:

Although Nūl is a standalone solution, it’s has to be deeply integrated with other data-generating and data-gathering solutions that a brand might use. Since most critical data for Nūl is internal—i.e. there’s more opportunity to optimize through understanding stores sales than search trends—Nūl’s efficacy would depend on how properly managed the data is.

Like most SaaS products, Nūl runs on a subscription. But despite targeting SMEs, it behaves like an enterprise solution in two key ways:

Still, the pricing remains flat and is billed monthly, much like other SME-focused subscriptions.

While Southeast Asia isn’t on China’s scale, it’s deeply woven into the fabric of global fashion. Thailand’s share has faded, but others—especially Vietnam—have surged. Taken together, mainland Southeast Asian countries now account for nearly 10% of global textile exports.

For many countries in the region, textiles aren’t just an industry—they’re the core of the export economy: 35% of total exports in Cambodia, 26.2% in Myanmar, and 18.8% in Vietnam. And Vietnam is the regional anchor, with 7,000+ textile factories, 3 million workers, and 80% of its output going abroad.

All of this adds up to one thing: inefficiencies here have ripple effects that go beyond Southeast Asia. Nūl wants to address those inefficiencies.

And there’s a lot of inefficiency to fix.

Globally, the fashion industry is responsible for 6–10% of carbon emissions. A single piece of apparel can emit up to 6 kg of CO2. And 30% of fashion products go unsold.

But those are global numbers. For Southeast Asia, the place to start is labor productivity. Because producing the wrong thing and doing it inefficiently is a double hit to the bottom line. In Vietnam, for example, productivity is 50% lower than in China. And since textile products have low value add, productivity in this industry is lower than manufacturing overall.

Thus, environmental issues are even more pronounced in this region. In Cambodia, textile-linked industries contribute 85% of industrial waste, 70% of toxic water pollution, and 27% of industrial air pollution. Vietnam is estimated to be responsible for 16% of global textile waste. Even if those numbers are off, the point remains: low productivity tends to equal more waste.

90% of Cambodian firms use only basic digital tools. In Vietnam, most companies still don’t use cloud computing, big data, or robotics—all under 7% penetration. Myanmar doesn’t have much data, but with factories experiencing 2.3 electricity outages per day, it’s safe to assume the tech adoption there isn’t great either.

So, most work is still labor-intensive. That’s part of the reason for low productivity. But it also means losses are worse: a well-maintained machine follows instructions. A human can make mistakes.

So we have a large industry, with low productivity, low tech adoption, and serious environmental consequences.

I think what is often ignored with data businesses is that all this optimization sounds great—until the reality of unstructured, irrelevant, and unreliable data smacks you in the face. The vast majority of companies, big or small, don’t know how to collect and manage their data. Until they figure that out, they can’t properly use solutions such as Nūl.

A large chunk of Southeast Asia’s garment production still runs on the CMP model (Cut, Make, Pack)—especially in Myanmar, Cambodia, and parts of Vietnam. These factories don’t manage inventory or demand—they just produce to order. That means Nūl’s product might not be applicable to the industry’s backbone.

AI needs time and data to figure out which signals matter—how much does a 7.4% uptick in unsold inventory matter? Or a rainy weekend? Until there’s enough data, the AI can’t optimize effectively. Smaller factories will take longer to teach the system what matters.

This one’s very speculative—but what are brands actually prioritizing? Probably brand-building, performance marketing, and expanding distribution. They likely know inventory’s being managed inefficiently, but is it high enough on their to-do list?

If local businesses are among the most inefficient, then this is the region where a solution such as Nūl can perform at its best. Moreover, if Nūl is able to overcome the challenges we’ve discussed and expand into more tech-ready regions, adoption should be much easier. Southeast Asia is where the company will have the most trouble integrating into companies’ workflows, but also where its product will bring the most value.

There’s a lot of noise around ESG, and some of the criticism is certainly valid. However, there’s no doubt that companies care about it. Combining improvements in ESG goal attainment with increased profitability is a very strong selling point in today’s world.

The region doesn’t have established institutions churning out data scientists or operations teams with strong math backgrounds to solve this problem on their own. Agentic AI is meant to be that team—sometimes augmenting an internal team, and in other cases replacing or creating one from scratch. That has real appeal in a resource-constrained region.

We have a lot of inefficiency in the world—and a lot of AI startups launching across the globe to tackle it. If even a fraction of them succeed, how much more efficient could the world become?