From Static to Dynamic: Dr. Ravi Changle Shares How Generative AI is Transforming Business Workflows

Dr. Ravi Changle, Director of AI and Emerging Technologies at Compunnel and a Forbes Technology Council member, recently spoke to Tech Achieve Media to share insights on the revolutionary impact of generative AI on business workflows. He notes that traditional static approaches are giving way to dynamic, AI-driven processes that unlock unprecedented efficiency and innovation. According to Dr. Changle, generative AI is transforming industries in more ways apart from just automating content creation, enhancing data analytics, and streamlining decision-making. He highlights how businesses can leverage these advancements to unlock new revenue streams, improve customer experiences, and stay ahead in a rapidly evolving market landscape. As a pioneer in AI adoption, Dr. Changle offers actionable advice on harnessing the power of generative AI to future-proof business operations and drive growth. 

TAM: What are the most significant disruptions Generative AI is currently bringing to the tech industry, and how do these changes impact both enterprises and consumers?

Dr Ravi Changle: While we are on the topic, let’s talk about generative physical AI.

Generative physical AI involves the integration of generative AI with physical AI. Physical AI refers to the mechanical aspects, including machines, robots, and similar technologies. By leveraging generative AI in this realm, we are moving towards collaborative bot frameworks, known as co-bots, instead of traditional bot frameworks.

Although collaborative bot frameworks are still in development, people have started working on them. Historically, we focused more on armed bots, but now we are creating robots with more advanced capabilities. Recently, our team filed a patent for this concept, and we are soon launching a generative physical AI-based collaborative bot. This bot can interact within the retail domain, understand its environment through cameras and IoT devices, and communicate using speech. This bot framework incorporates a layer of large language model operations, enabling it to interact more intelligently. Essentially, it combines robotics and bot frameworks, representing the next wave of generative physical AI.

In terms of applications, this technology will assist in industrial control systems, inspection of manufacturing units, and operations in farmlands. For example, we have developed intellectual property (IP) for a platform that includes a mobile application, a web application, and a robot. This robot can visit farms, capture insights, take pictures, and diagnose plant diseases. It can then recommend necessary treatments and identify nearby vendors supplying the required pesticides or medicines. In manufacturing units, the same technology can manage inventory, analyze pilferage, and detect any alarming issues. Predictive maintenance will also advance significantly, contributing to the evolution of industry 5.0 maturity.

TAM: How can businesses ensure that their use of Generative AI yields insightful and actionable outcomes rather than generic insights?

Dr Ravi Changle: There are two main approaches to this: a strategic perspective and an operational perspective.

From a strategic standpoint, proper governance is essential to prevent deviations. Governance means establishing mechanisms that transform strategic plans into operational activities. To achieve this, we need to set up Key Performance Indicators (KPIs) from both the subject matter and functional aspects of the application, as well as business KPIs.

Aligning these KPIs and tracking the output generated by the generative engine helps prevent hallucinations. If a hallucination occurs, the generative AI engine will identify it, provide a score, and request human intervention. We have already implemented this mechanism for our projects and external clients, using a metric-based evaluation system for monitoring.

Through this monitoring, we can detect hallucinations and determine if the model requires further fine-tuning. We can apply a reinforcement learning layer to our custom data. This process involves more than just running the generative engine in the background; it requires an RLHF (Reinforcement Learning from Human Feedback) framework for our custom data.

The RLHF framework includes a reward and penalty mechanism, allowing the system to automatically adjust based on errors and accuracies. This AI agent helps safeguard against failures, ensuring the engine provides accurate responses.

TAM: Can you explain how generative AI is shifting business operations from static workflows to more adaptive systems? How do adaptive AI systems handle unexpected changes in business environments compared to traditional static workflows?

Dr Ravi Changle: When discussing static versus adaptive frameworks, it’s important to recognize that initially, we believed these frameworks would be primarily suitable for autonomous tasks and not very creative in generating responses. An adaptive framework, however, involves understanding the environment, learning from it, and then responding or reacting accordingly. The distinction between a response and a reaction lies in the level of proactivity and aggressiveness: a proactive, aggressive response becomes a reaction.

We can now implement these adaptive frameworks using generative AI capabilities. The approach varies by industry and use case, but there are mechanisms to enhance adaptability. One such method involves using a reinforcement learning-based RLHF (Reinforcement Learning from Human Feedback) layer, incorporating human feedback to create adaptive systems.

For example, imagine an agent that scans documents and generates insights. Document automation, once based on natural language processing, has evolved with generative AI maturity models. Subject matter experts, such as scientists, can use this technology for ongoing projects, with the GNI engine scanning requirements and project details, integrating agile practices, and providing outputs tailored to research needs.

The GNI bot can identify research gaps, suggest potential interactions (e.g., between molecules in drug discovery), and recommend next steps to address previous research gaps. Subject matter experts, such as doctors or manufacturing supervisors, can customize the system through prompt engineering, making it more adaptive and contextually aware.

Adaptive AI systems handle unexpected changes better than traditional static workflows, which rely on predefined rules and processes and are difficult to adapt. Adaptive AI continuously learns from new data, identifying uncertainties (or risks) and modeling them within the predictive environment. For instance, in a volatile business scenario, an adaptive AI system can identify risk variables, incorporate them into a risk management framework, and provide insights into risk exposure. This can apply to various risks, including enterprise, market, and product failure risks.

Adaptive AI can use both quantitative and qualitative measures to detect anomalies and predict outcomes, offering a more comprehensive and dynamic approach to decision-making. It learns from the environment and provides insights into risk exposure, helping you decide whether to proceed with certain actions.

TAM: What challenges do companies face when integrating AI into their strategic planning processes? 

Dr Ravi Changle: There are some strategic challenges involved, particularly from a long-term vision perspective. These include setting up a governance mechanism and identifying potential use cases to focus on, which presents the first dilemma.

Many organizations and industries have yet to determine how exactly AI can be leveraged within their ecosystems. As part of our services, we offer consulting on identifying potential ROIs attached to specific projects, which is a more strategic endeavor. This requires brainstorming sessions with CXOs and upper business units.

Regarding functional aspects, business functions face challenges integrating AI or generative AI-based mechanisms into their ecosystems. These challenges could stem from data integration issues, the need for third-party data connectors, or existing ERP systems. Customizing these integrations presents functional unit challenges.

Operationally, automation is another aspect to address, where confusion often arises between workflow automation and generative AI-based automation. Raising awareness about how RPA (Robotic Process Automation) and generative AI can work together or within a hybrid system is essential. One major operational challenge is the skill gap. Not every industry has the required skill sets for these advancements. For instance, large language model operations require MLOps engineers who understand CI/CD pipelines, how to monitor generative AI agents, prevent hallucinations, and compare bot outputs with human subject matter experts. Applying a reinforcement learning layer to automate these processes is crucial.

Other constraints include budget, time, and the speed of automation. Many industries and business offerings are rapidly evolving. For example, web design has become prompt-oriented, with tools like Figma making design easier. Instead of fearing job displacement, it’s important for everyone to learn how to integrate generative AI into their current workflows and upskill themselves.

TAM: Why is strategic AI leadership crucial for business transformation, and how does the role of a Chief AI Officer fit into this?

Dr Ravi Changle: When we talk about the role of a Chief AI Officer (CAIO) or offering Chief AI Officer as a Service, it’s not just about an individual. It’s a council that includes the CAIO, an AI council, a steering committee, and stewardship. Together, they envision how to integrate business units with technological adoptions and adaptations to address the challenges of rapid global growth.

The CAIO and the council are responsible for setting a strategic pathway for business transformations over the next 20 years. With the advent of Industry 5.0, we will see significant disruptions across economies. For example, collaborative bot frameworks and machine customers will change how we interact. We will negotiate with machines that make buying decisions, much like we currently deal with autonomous sellers in dropshipping environments.

This scenario requires visionary leadership from the CAIO, who will guide the business use cases and future strategies to maximize shareholder wealth, ensure business growth, and maintain sustainability. These leaders will establish strategic units to oversee all business functions, including logistics, supply chain, retail, marketing, production, and manufacturing, aligning them with customer demands.

In creative fields like photography and filmmaking, platforms like Sora will enable professionals to create high-quality content quickly, blending fiction and reality. In scientific fields, tools like GitHub Copilot can expedite development processes, saving time, money, and effort.

Planning for global AI adaptation is crucial, as it impacts society positively and negatively. Leveraging AI for social good is essential; if AI complicates our lives, it defeats its purpose. Therefore, visionary leaders must align societal needs with best-in-class AI practices for the benefit of customers, stakeholders, governments, and financial systems, ensuring continuous progress.

TAM: What are the most significant ways in which General AI is expected to transform different industry sectors? Can you highlight some innovative applications of General AI beyond conventional automation and data analysis?

Dr Ravi Changle: We have leveraged generative AI alongside general AI principles to devise a framework that reduces digital waste. We’re working with our clients to identify mechanisms to scan entire cloud platforms, assess data assets, compute usage, and track development efforts, time, and money.

By pinpointing idle machines and anomalies, we can recommend reductions in digital waste, translating into significant savings. These savings can then be reinvested into R&D to adopt best practices. It’s clear that generative AI will become an integral part of future services, and organizations without generative AI principles in place will struggle to remain sustainable in the next few years.

The first point I want to address is the necessity of generative AI for long-term sustainability. The second is optimization. As mentioned, there are various ways to optimize the usage of generic AI within our ecosystem. Reducing digital waste also means cutting down on wasted time, money, and effort.

Strategic efforts are needed to identify problem complexities and internal capabilities. By focusing on these metrics, we can pinpoint golden opportunities and leverage internal capabilities over third-party solutions, thus upskilling our teams. This approach allows us to maximize AI’s potential for automation and recommendations, which were previously difficult due to human limitations and diverse opinions. Generative AI, unlike human hierarchies, does not adhere to any hierarchy or designation, making it an unbiased tool for optimization.

TAM: How is Compunnel positioned to help businesses navigate the complexities of the Generative AI ecosystem? What can we expect from you on this front?

Dr Ravi Changle: I’d like to highlight our achievements in generative AI, AI, and cybersecurity. When I joined the Forbes Technology Council, we already had numerous clients in predictive maintenance, industrial IoT, banking, and pharmaceuticals, all employing our generic AI practices.

In the 2020s, we leveraged GPT models and other large language models to identify the next wave of AI. We quickly capitalized on this for our clients, integrating AI into their existing businesses. Komperol has a strong track record with decade-long client relationships, providing AI components in application development, web frameworks, and digital solutions. This has led to significant returns on investment and boosted our confidence.

Within a year, we developed 40+ accelerators and showcased them on major platforms. We are also preparing to globally exhibit from the UN AI for Good perspective. I represent Compunnel in various councils and recently discussed AI resilience with industry leaders.We are developing a global AI cybersecurity task force to safeguard societal and environmental interests, ensuring green AI implementations. Our initiatives include automated data engineers for clients to migrate ecosystems, integrating Data Vault 2.0 within Microsoft Fabric, and steering client economies, including government collaborations.

Our EFG model focuses on AI for Good, helping in education, farming, and government sectors. We’ve tackled diverse use cases, from farming equipment manufacturers to aerospace engineering. We are currently working on a model for which we will soon be obtaining a patent. This model features Wi-Fi enablement, GPS, sensors, and various other advanced components. It even includes the Jetson Nano, which helps us prompt and engineer responses effectively.

In working on collaborative bots, we can develop them on the cloud, workstations, or servers. Our capabilities extend to robotic models, environmental models, and simulation-ready assets. For example, platforms like Siemens’ simulation platform and Autodesk’s Fusion platform allow us to work with models, sensors, utilities, synthetic data, and artificial data sets.

We can train these models and robots in simulation environments and then apply them to various industries such as energy, retail, and real estate. Essentially, we are focusing on custom robotics and simulation solutions, leveraging these technologies behind the scenes.