AI adoption shifts as enterprises move from open APIs to offline, proprietary models: M37Labs

As Artificial Intelligence (AI) adoption grows in India, businesses are grappling with challenges like return of investment (RoI) uncertainty, scalability issues, and vendor lock-in. Despite the demand for automation and efficiency, many enterprises struggle to implement AI effectively. 

M37Labs, founded by Prashant Iyer (CEO) and Zorawar Purohit (CAIO), focuses on AI consulting and product development. The company works with enterprises to integrate AI through its AI Ignition framework and builds modular systems that operate across different platforms, avoiding reliance on a single vendor. 

In a conversation with TechCircle, Iyer and Purohit break down the key barriers to AI adoption, how companies are addressing them, and what’s next for AI in enterprise technology. Edited excerpts:

What are the main challenges companies in India face when scaling AI, and how do you help them overcome these challenges? 

Iyer: Over the past five months, we've worked with companies of all sizes, and there's a clear understanding that AI can enhance operations. However, some C-suite executives hesitate due to concerns about RoI. 

To address this, we've developed the AI Ignition framework. Many executives are eager to integrate AI into their strategies, signaling strong interest. While global reports show slower AI adoption, we’ve seen openness and meaningful discussions here, expecting broader adoption in the next 3-6 months. 

From our conversations with C-suite leaders across industries like banking, financial services and insurance (BFSI), retail, and media, it’s clear that AI can significantly boost efficiency. Many companies still rely on manual processes, like a major retailer we met recently, where supply chain and procurement are entirely manual. The key question is: how can AI streamline these processes and improve efficiency quickly, often within 60 days? 

AI can drive transformation fast, improving workflows, productivity, and RoI. This is just a glimpse of the AI opportunities we’re seeing. 

Purohit: The Software-as-a-Service (SaaS) ecosystem has been around for decades, with challenges like vendor lock-in and scaling in a fast-evolving landscape. The real issue now is the rapid emergence of AI models like GPT-4.0 and DeepSeek R1. Enterprises struggle with choosing the right one, and without a flexible system, they face vendor lock-in and miss out on advancements. 

Our solution is a customised approach, building systems with modular design. We use both open and closed-source ecosystems, writing tailored code for each enterprise, ensuring efficiency today and adaptability for future AI developments. 

How is your company positioning itself as an alternative to big tech, and what makes your approach faster and more cost-effective for AI adoption? 

Iyer: AI Ignition is generating significant interest and traction. This is where we bring together our expertise in business strategy, consulting, and technology. 

We collaborate directly with the C-suite to identify key challenges, define expected outcomes, and deliver AI solutions fast. Unlike traditional big-four consulting, which involves long engagement timelines, our approach accelerates the entire process. We streamline everything from design thinking to rapid prototyping and deployment, reducing the timeline to just 60 days. This faster time-to-market unlocks major efficiencies for enterprises, making AI adoption more seamless and impactful. 

Purohit: Many big players follow the DVF (Desirability, Viability, Feasibility) framework. They’ve been around longer, but as you know, the bigger the organisation, the slower the shift. We were born in the AI era, and our teams are deeply immersed in it. Our partners aren’t just PhDs, they have hands-on experience developing enterprise AI long before the GPT boom. We entered the market at the right time, just as AI began gaining global traction, and that expertise has been a clear differentiator. 

We are more cost-efficient, and we bring all the expertise in-house while understanding how large enterprises operate. This combination results in faster DVF assessments, reduced time to market, and lower costs. In short, we deliver highly accurate AI consulting, rapid prototyping, and clear go/no-go decisions for leadership at a significantly lower price point than competitors, without compromising on AIDT design thinking, prototyping, or expertise. 

We haven’t marketed it this way, maybe we should. But billion-dollar companies and other clients we’ve engaged with recognise this advantage immediately. 

AI has evolved from traditional automation to an intelligent, adaptive ecosystem. From your perspective, how has this shift happened? What key AI characteristics are driving enterprises today? 

Purohit: AI capabilities are evolving rapidly. Take the latest 4.0 model, if you've seen the news, you'll know that as of yesterday, it now includes image generation. While tools like Stable Diffusion, DALL·E, and Midjourney have been around for a while, they’ve had limitations.  

AI's evolution extends beyond image generation. Machine learning, a subset of AI, has been around since Google’s early days. It’s why search results have been so accurate for decades. It also powers things like Netflix’s adaptive streaming, which adjusts video quality based on internet speed by subtly reducing pixel detail, something the human eye can’t detect. So, while we think we're watching the same movie, machine learning is working behind the scenes to optimise the experience. 

Initially, AI was limited to a small group of engineers and enterprise applications. This mirrors other technological advancements, like GPS. As Greg Milner explains in Pinpoint, GPS was originally designed for military precision, and even today, civilian GPS is less accurate than the military’s reserved version. 

AI is following the same trajectory, starting as a niche technology, gradually becoming more accessible, open-sourced, and democratised. As more brilliant minds contribute, its applications expand across industries, continuously reshaping its characteristics and capabilities.

Which industries do you think AI is expanding rapidly in? Specifically, in the BFSI sector, AI started with chatbots for customer experience but has evolved to support internal workflows. What’s your take on this shift? 

Purohit: AI has been around longer than most realise. The surge in AI adoption, especially with chatbots, is largely due to ChatGPT's rise during Covid. The first place AI was applied was where communication is key, like BFSI, which has strict compliance and zero-error requirements. Similarly, AI has long been used in aviation systems with no tolerance for mistakes. 

BFSI adopted AI because of the massive rise in digital transactions, making it necessary to either hire thousands of people or use AI to handle the volume. They also have the capital and need, which is why they started with customer support and are now focusing on fraud detection and similar tasks. 

While sectors like retail are adopting AI slower, big changes are coming. For example, AI could soon summarise meeting notes and handle transactions with voice commands instead of typing. In retail, you might even try a car using augmented reality from home. 

So, while BFSI is leading the way, industries everywhere are building systems that will soon revolutionise how we interact with technology and streamline workflows. 

How does your company see AI agents integrating with traditional business structures? And what are your thoughts on job loss concerns due to AI agents? 

Iyer: Our vision for the future is that AI will significantly boost workforce productivity. For example, imagine a company with a revenue of ₹100 crore. By implementing an AI-driven system, that company could achieve 10 times the productivity with the same number of employees, potentially growing to ₹1,000 crore. This is the larger shift we see happening across industries globally. 

AI will enhance productivity, driving both business growth and profitability, while also increasing income for employees. We view this as a positive multiplier effect. However, similar to past tech cycles, some industries may see certain roles or processes become redundant. For instance, in creative services, roles like the person responsible for transferring art negatives to printers became obsolete with the advent of CDs. 

With AI, though, the focus is on augmentation. Each person within a company will be able to leverage AI to handle much more work, boosting overall productivity. This will create a more efficient and effective workforce, enabling companies to achieve much greater results. That's how we see the future unfolding. 

What are your company’s top tech priorities for the next 12-18 months, and how do you plan to stay ahead in the evolving AI landscape?

Iyer: As we've gained experience, we've realised that while the services consulting model will continue to thrive in the coming years, we also have a unique opportunity to develop specialised products in certain areas. We're already working on a few of these. Our plan is to combine our AI consulting services, which will drive enterprise AI adoption, with AI products that are built in India but deployed globally. Over the next 12 to 18 months, our focus is to build teams that excel in both AI consulting and creating top-notch AI products for a global market. 

Purohit: Our edge lies in the evolving surf production model. Right now, it’s agentic, but eventually, it will move into Brain-Computer Interfaces (BCI), and we’re also seeing early signs of quantum computing. Machine learning and AI will continue to expand beyond current use cases. 

The future is about proprietary data and models. We’ve moved from open APIs to offline models running on my MacBook, enabling rapid prototyping without large enterprise costs. We can deliver solutions with enterprise-grade security, combining open-source models for low-risk decisions and enterprise models for high-risk ones. 

We’ve identified patterns across industries. For instance, in e-commerce, we built a convolutional neural network to automate metadata generation, solving a common issue for companies with large product catalogs. This is a key opportunity for both service and product development. 

In the next 12-18 months, our blend of service expertise and product development will allow us to stay ahead and drive industry change. 

---