Turning AI from buzzword to backbone

In conversation with Sujit Janardanan, Chief Marketing Officer – Neysa

Sujit Janardanan has spent over 20 years building and leading marketing functions for some of the most respected names in the technology space, from Google Cloud and AWS to Cropin and Netmagic. He understands what it takes to bring clarity and direction in a fast-changing landscape. Now at Neysa, he’s helping companies make smarter, faster decisions using technology that’s actually built for them. 

In this conversation, Sujit shares grounded, real-world insights on how pharma businesses can solve everyday problems, whether it’s keeping machines running smoothly, improving quality checks, or managing compliance – not with jargons, but with practical solutions that work.

Predictive maintenance is becoming a big buzzword in pharma manufacturing. In your experience, how is AI helping to keep machines running longer and cutting down on unexpected breakdowns? 

Predictive maintenance is far more than a buzzword; it’s the foundation of the ‘lights-out’ or smart factory of the future. In essence, AI gives your manufacturing equipment a voice and the ability to tell you it’s feeling unwell, long before it gets sick. 

It works by deploying sensors on critical machinery to capture real-time data, like thermal signatures, vibration patterns, and acoustic anomalies. AI models, running on a high-performance cloud infrastructure, are trained on this data to learn the unique signature of ‘healthy’ operation.

They can then detect minuscule deviations that are precursors to failure. 

The impact is transformative. Instead of reacting to a costly breakdown that halts a production line and potentially compromises an entire batch, manufacturers can shift to a proactive, data-driven maintenance schedule. This leads directly to: 

• Increased Overall Equipment Effectiveness (OEE): Maximizing uptime and output. 

• Reduced Spoilage: Preventing equipment malfunctions that can ruin sensitive, high-value batches. 
• Optimized Costs: Minimizing emergency repair costs and extending the operational life of expensive machinery. 

Ultimately, predictive maintenance turns manufacturing from a reactive process into a predictable, self-aware system, which is a cornerstone of modern GXP (Good ‘x’ Practice) compliance. 

When it comes to checking product quality, whether it is tablets, vials, or blister packs, how does AI-powered visual inspection stack up against traditional manual checks? 

If predictive maintenance gives machines a voice, AI-powered visual inspection gives them superhuman sight. It represents a quantum leap in accuracy, consistency, and speed over traditional manual checks. 

Manual inspection, while essential, is inherently limited by human subjectivity and fatigue. An inspector can have a bad day; an AI model does not. Here’s how AI stacks up: 

• Precision: AI vision systems can detect microscopic defects, subtle cracks in a vial, incorrect tablet embossing, or minute foreign particles, that are invisible to the naked eye. 
• Consistency: The AI applies the same criteria to the first unit and the millionth unit, 24/7, eliminating the variability that comes with human inspectors working across different shifts. 
• Speed & Data Integration: An AI system can inspect thousands of units per minute, a scale humans simply cannot match. More importantly, every inspection generates data for correlation with the predictive maintenance system. For instance, we can identify that a slight increase in machine vibration is causing a 0.5% increase in cosmetic defects on a blister pack. And this can create a powerful feedback loop for total quality management that was previously impossible. 

For a regulated industry, this moves quality control from a probabilistic sampling method to a deterministic, 100% inspection model, creating an unparalleled audit trail. 

High-performance computing and advanced GPUs are making serious waves in pharma R&D. How are they speeding up things like molecular modeling, process simulations, or even generative formulation design? 

You’re right, and this is where the entire value chain truly begins. The impact of HPC and GPUs in R&D is nothing short of revolutionary. It’s collapsing the R&D timeline from decades to years by shifting the paradigm from slow, physical experimentation (in-vitro) to rapid, computational discovery (in-silico). Here’s how: 

• Accelerated Molecular Modeling: Simulating how a single drug candidate molecule will dock with a target protein is a massively complex computational problem. With advanced GPUs, researchers can screen millions of potential molecules virtually in the time it used to take to analyze a few hundred in a lab. 
• Generative AI for Drug Discovery: This is the next frontier. Instead of just screening existing libraries, generative AI models can now design entirely novel molecules and formulations optimized for specific properties like efficacy, stability, and low toxicity. 
• Process and Scale-Up Simulations: Before a single gram is produced, AI can simulate the entire manufacturing process. This helps scientists predict how a formulation will behave in large-scale bioreactors, optimizing for yield and stability and dramatically reducing the risks associated with technology transfer from lab to plant. 

This requires a specialized “AI-native” cloud infrastructure system, like Neysa’s, which is purpose-built with the raw GPU power and high-speed data handling needed to make in-silico R&D a reality. 

We often hear about open-source and modular AI setups. For a highly regulated industry like pharma, how does that kind of flexibility help with compliance and avoiding vendor lock-in? 

This is a critical point. For many industries, a “black box” AI solution is acceptable if it delivers results. For pharma, it’s a non-starter. The ability to validate every step of a process is paramount, and this is where open-source and modularity become strategic assets. 

• Compliance and Validation: A modular, open-source approach allows pharma companies to “look under the hood.” They can dissect, understand, and independently validate each component of their AI workflow. This transparency is essential for satisfying regulatory bodies like the FDA or EMA, who require clear explanations of how a decision was made (so-called “explainable AI” or XAI). 
• Avoiding Vendor Lock-in: The pharmaceutical landscape changes constantly. Relying on a single vendor’s proprietary, monolithic system is risky. A modular setup allows a company to use the ‘best-of-breed’ tool for each specific task, be it for data ingestion, a specific modeling algorithm, or visualization, and swap components out as better technology emerges, without having to rip and replace the entire system. 
• Future-Proofing: This flexibility ensures that as regulatory requirements evolve, the company can adapt its AI workflows accordingly, rather than waiting for a single vendor to issue a patch. 

Security and transparency in AI systems are becoming just as important as the results they deliver. How can pharma manufacturers build AI workflows that are not only effective but also trustworthy and audit-ready? 

Trust is the currency of the pharmaceutical industry. An AI system that is not secure, transparent, and auditable is not just ineffective; it’s a liability. Building trustworthy AI is a foundational requirement, not an optional extra. 

The key is to design the infrastructure and workflows with an “audit-ready by default” mindset. This involves three pillars: 

• Data Lineage and Integrity: You must be able to trace every piece of data from its point of origin through every transformation and analysis step. And create an unbroken chain of custody, ensuring that the data feeding your AI models is secure and untampered with. 
• Model Traceability: It’s not enough to get the correct answer; you need to prove how you got it. That means meticulous version control for AI models, logging which model version was used to analyze which batch of data at what specific time, and documenting its performance metrics. 
• Robust Security Posture: The underlying infrastructure must provide end-to-end encryption (for data at rest and in transit), stringent access controls, and continuous monitoring to protect sensitive intellectual property and patient data. 

At Neysa, we build our AI Acceleration Cloud system on these principles, ensuring our pharma clients have the “glass box” they need, where the process of getting an answer is just as straightforward and defensible as the answer itself. 

Neysa’s AI-native cloud comes with flexible, pay-as-you-go pricing. For pharma companies that need to balance innovation with tight budgets, how does this approach make adoption easier? 

The single most significant barrier to AI adoption isn’t a lack of ideas; it’s often the perceived risk and the massive upfront capital expenditure (CapEx) required for specialized computing infrastructure. Our pay-as-you-go model directly dismantles this barrier. 

• It effectively democratizes access to cutting-edge AI. 
• Eliminates Upfront CapEx: A company doesn’t need to spend millions on an in-house GPU cluster. They can access world-class infrastructure immediately and pay only for the compute hours they use. 
• Enables a “Crawl, Walk, Run” Approach: A pharma company can start small with a pilot project, perhaps for visual inspection on a single production line. They can prove the ROI with a manageable operational expense (OpEx), build a business case, and then seamlessly scale up to other lines or into R&D without needing a new budget cycle for hardware procurement. 
• Aligns Cost with Value: This model aligns cost directly with innovation and experimentation. If an R&D project requires a massive burst of computation for a week, you pay for that week. If a manufacturing line runs 24/7, you pay for that consistent load. This financial agility is critical for balancing tight budgets with the urgent need to innovate. 

By being a homegrown AI infrastructure provider, how does Neysa give pharmaceutical manufacturers an edge compared to relying on global cloud giants? 

While global clouds are powerful, partnering with a homegrown, AI-native provider like Neysa offers distinct strategic advantages, particularly for the Indian pharmaceutical industry, which is a worldwide powerhouse. 

• The primary advantage hinges on sovereignty and specialized partnership. 
• Data Sovereignty: For a pharma company, its data, formulation IP, clinical trial results, and manufacturing parameters are its crown jewels. Using a homegrown cloud ensures this sensitive data remains within India’s borders, simplifying data governance and aligning with national data privacy and security mandates. 
• Local Regulatory Understanding: We work with partners who have a deep, on-the-ground understanding of the Indian pharma landscape, including the specific requirements of bodies like the CDSCO. Our solutions and support are tailored to this context, not a one-size-fits-all global template. 
• Performance and Cost: By hosting infrastructure locally, we significantly reduce latency, which is critical for real-time applications like robotic control or high-speed visual inspection. Furthermore, our pricing is more transparent, without the complex and often costly “data egress” fees that global players charge for moving data. 
• True Partnership: We are not just a utility provider; we are part of the same ecosystem. We provide hands-on, local support to help our clients build and deploy their AI solutions, acting as a true strategic partner in their digital transformation journey. 

Looking ahead, as AI takes on a bigger role in drug discovery and smart manufacturing, how do you see AI platforms evolving to meet pharma’s unique mix of innovation demands and strict regulatory standards? 

Looking ahead, the evolution of AI Systems will be defined by one word: convergence. Today, we often talk about AI in R&D and AI in manufacturing as separate domains. The future is an integrated, end-to-end intelligent platform that creates a continuous feedback loop across the entire pharma value chain. 

AI platforms will evolve to become the central nervous system of the pharmaceutical enterprise, and they must: 

• Unify Diverse Data: They will need to seamlessly ingest, process, and correlate incredibly diverse data types, from genomic data in early discovery, to real-time sensor data from bioreactors, to quality control imagery, and even real-world evidence from patient outcomes. 
• Enable the “Digital Twin”: Platforms will host comprehensive ‘digital twins’ of both the product and the process, allowing companies to simulate the entire lifecycle of a drug, from how a formulation change might impact manufacturing yield to how it might affect patient bioavailability. 
• Embed Compliance by Design: As complexity grows, compliance cannot be an overlay. Future platforms will have regulatory rules and GXP principles embedded into their very architecture, automating documentation and making audit trails an intrinsic part of every operation. 

At Neysa, this is our vision. We are building the foundational platform for this converged future, an infrastructure that is powerful for accelerating next-generation drug discovery, agile enough for smart manufacturing, and trustworthy enough for the most stringent regulatory demands. The goal is to empower the pharmaceutical industry to bring safer, more effective medicines to market faster than ever before.