From Tasks to Agents: Redefining Business Process Automation with Tuhin Chakraborty

Understanding Task Mining: A New Lens on Productivity

Task mining represents a significant leap forward in how organizations can understand and improve their operational efficiency. Imagine trying to identify an obscure workflow that is silently draining millions from your budget. Traditional methods often fall short, but task mining, powered by Artificial Intelligence (AI), offers a path to clarity and actionable insights. This technology works by making inferences about the work people do on their computers using AI. Essentially, it records user interactions like clicks and keystrokes, and then applies advanced Machine Learning algorithms to process this data. The goal is to determine what tasks and processes individuals are engaged in, where their time is predominantly spent, and, crucially, where the most promising opportunities for Automation and process improvement lie. The output often includes detailed process maps that serve as a foundation for strategic Digital Transformation initiatives.

From a client’s perspective, task mining provides a previously unattainable level of visibility into their own operations. It moves beyond anecdotal evidence or manual observation, offering a data-driven view of how work actually gets done. This clarity is the first step towards meaningful Business Process Management (BPM) enhancements and impactful AI-driven savings.

Real-World Impact: Uncovering Hidden Inefficiencies

And one of the key insights that Mimica was able to identify for this customer is that claims involving a tow truck, so these are auto claims, so you know, when you have a car accident, you file a claim, claims involving a tow truck take 30% longer to process than claims that don’t.

The practical applications of task mining are vast, particularly in industries with complex, high-volume processes. Consider the insurance sector, where a major US company utilized this approach to dissect its claims processing operations. This company was spending a staggering $240 million annually on claims, yet lacked a detailed understanding of how this money was being utilized across various activities. They couldn’t pinpoint where process improvements or Automation could yield the best results.

By recording the work of 1,000 claims agents, the AI was able to identify all the distinct activities involved in the claims process. This allowed the company to establish productivity benchmarks for each activity, effectively mapping out how that $240 million was distributed. One of the most striking discoveries, as Tuhin Chakraborty shared, was related to auto claims involving tow trucks. The root cause wasn’t immediately obvious, but further investigation revealed that the case management system’s workflow for tow-truck-related claims was significantly less efficient, requiring more clicks and keystrokes.

This new understanding, derived directly from task mining data, was categorized by the company as a new form of business intelligence. It empowered the product manager responsible for the case management system to redesign the inefficient workflow. The result was a remarkable 80% reduction in the processing time difference, leading to approximately $5 million in savings. This success story underscores how task mining can turn messy clickstreams into crystal clear process maps, converting AI hype into tangible, CFO-level ROI, a core challenge in many Digital Transformation journeys.

Addressing Privacy in Human-Centered AI

We actually put them at the center of the recording experience, and the reason is that every employee and really every person has something that they would rather be doing with their time.

The prospect of recording employee computer activity naturally raises questions about privacy. However, effective task mining solutions are built with a Human-Centered AI approach, placing employees as active participants rather than passive subjects. The system isn’t a hidden background recorder; employees are central to the recording experience. This is driven by the understanding that most employees have repetitive, mundane tasks they would prefer to offload, allowing them to focus on more engaging and core responsibilities.

Employees are often eager to participate because task mining can identify these burdensome tasks for Automation, freeing up their time for more valuable work. To ensure privacy and foster Trust, several controls are implemented. The recorder is always visible on the desktop, not hidden. Users have complete control with start, stop, and pause buttons, allowing them to halt recording if they are performing personal tasks. Perhaps most critically, all information processed is anonymized. Confidential or personally identifiable information, such as passwords, names, emails, and addresses, is never recorded or is stripped out, ensuring that the resulting process maps do not contain sensitive data. This focus on privacy and employee control is vital for ethical and successful AI adoption.

From Manual Mapping to Automated Discovery: The Evolution of Process Understanding

And he was emotional about the fact that for the first time in 15 years, this work would be taken off his plate.

Before automated tools like task mining, understanding business processes involved business analysts conducting interviews and manually drawing diagrams. This traditional approach, while valuable, often suffered from subjective viewpoints, where each stakeholder might perceive inefficiencies differently. Task mining offers a more objective reality, grounded in data.

The transparency provided by data-driven process discovery can be personally impactful for employees. In one instance, at a large telecommunications company, an employee became emotional upon seeing the results of a task mining analysis. The AI had identified an Automation opportunity for a process he had been performing manually for an hour every single day for 15 years: breaking down a large spreadsheet of recruitment candidates into smaller ones for distribution. For this individual, the prospect of automating this repetitive task was a profound relief, a gift that would free up a significant portion of his work life. This highlights the positive side of Employee Engagement when Automation is framed as a way to eliminate drudgery.

The journey from identifying the “as-is” process to defining the “to-be” state is transformed by this technology. Traditional methods rely on interviews and anecdotal evidence, which can be inaccurate due to misremembering, mischaracterizing inefficiencies, or biases. Task mining, however, provides a source of truth based on actual activity data. This accurate information about tasks, timings, and process maps forms a solid foundation for any transformation. Modern tools even allow for modeling different scenarios – for instance, assessing the impact of deploying Robotic Process Automation (RPA) or more advanced technologies like agentic AI to optimize or eliminate steps and visualize the future “to-be” process. This data-centric approach to Business Process Management (BPM) is critical for successful Digital Transformation.

The Symbiotic Future: Task Mining and Agentic Automation

In our view of the future agentic automation, you will not require a specialized skillset in order to build the automation.

Task mining and the emerging field of agentic Automation are inextricably linked. The vision for Mimica AI, from its inception, was a new type of Automation – one that doesn’t require explicit, click-by-click programming but instead learns by observation, much like humans do. To achieve this, the foundational step is the ability to accurately interpret work by recording and deriving meaning from clicks and keystrokes. The analytics tools available today are a stepping stone towards this broader vision of intelligent Automation.

But what exactly is agentic Automation from a business leader’s perspective? In essence, as Tuhin Chakraborty explains, it’s an automation approach that doesn’t necessitate specialized development skills. Traditionally, creating automation required developers trained in specific coding languages or RPA platforms. This reliance on a limited pool of specialists often bottlenecks the potential value of Automation. Agentic AI aims to democratize automation. In this future, any employee could potentially deploy an automation simply by demonstrating their process to an AI agent, which would then learn and reproduce it.

Consider the “before and after.” Previously, a business analyst would interview users, manually create process maps, and then hand these to an RPA developer for manual automation building – a process that could take six months or more. In an agentic AI world powered by task mining, the user performing the process, along with a recorder, would capture the workflow, which then automatically powers the automation. This is essentially automating the automation process itself, a significant leap in agility and efficiency.

A key challenge in this vision is capturing and automating the decision-making aspects of a process, the “thinking” that isn’t always visible on screen. For rules-based decisions (e.g., “if an invoice is over $50,000, it needs extra approval”), the belief is that AI can learn these rules by observing patterns in data over many instances. By correlating chosen paths with underlying process data, the system can derive decision rules. As Large Language Models (LLMs) and other AI technologies advance in their reasoning capabilities, they will be able to handle increasingly complex decisions, interpreting unstructured data from various sources to determine the right course of action. This continuous improvement in AI’s cognitive abilities is central to the evolution of Human-AI Collaboration in the workplace.

Orchestrating the Future: Humans and AI Agents in Partnership

The role of task mining in this evolving landscape is twofold. Firstly, it acts as an analytical layer across all digital work within an organization, aiming to make that work more efficient. It helps identify which human-performed tasks are prime candidates to be transitioned to AI agents. As the work humans do becomes more complex and creative, the analytics provided will also become more sophisticated, offering insights into optimizing this higher-value work.

Secondly, the ambition extends to owning the agentic layer itself and managing the transition of work between humans and AI agents. The future enterprise will likely feature an army of human workers collaborating with an army of AI agents. Task mining platforms aim to be the infrastructure layer orchestrating this Human-AI Collaboration, continually shifting appropriate work from humans to agents over time. While a world where AI does all the work and humans relax on a beach is a distant, perhaps even fantastical, notion, the journey involves progressively more sophisticated Automation.

Practical First Steps for Business Leaders in the Age of AI

The first step to take is really to do a holistic assessment of all the manual work happening in your organization to determine what percentage of that work and which parts of it can be given over to an LLM or to an agent that can begin automating that, that work.

For senior managers, department heads, and team leads navigating the complexities of Digital Transformation, the advent of LLMs and advanced AI like ChatGPT has undeniably unlocked new potentials for Automation and organizational efficiency. The critical question for Leadership is not *whether* this technology can transform their business, but *how* to implement it effectively.

The recommended first step, as highlighted by Tuhin Chakraborty, involves a comprehensive assessment of manual work. This involves determining what percentage of this work, and which specific parts, can be handed over to AI agents or LLMs. Tools specializing in task mining can facilitate this by providing a comprehensive understanding of manual processes quickly, allowing businesses to map, capture, and ultimately optimize them. Such an assessment provides a clear roadmap from the current state to a future vision where AI handles a significant portion of automatable tasks.

Gaining clarity and visibility into all work processes is paramount. Once this understanding is established, the path to transformation often becomes self-evident. Leaders will be able to distinguish which tasks can be automated “as-is” and which will continue to require human intellect and complex decision-making. The immediate future likely involves a hybrid workforce where humans and AI agents work alongside each other. Humans will focus on tasks demanding intelligence and intricate judgment, while AI agents handle the repetitive, mundane aspects. The leader’s role is to understand their processes in sufficient detail to identify where these handoffs between human and AI should occur and to design this new, hybrid operational model. This thoughtful integration is key to harnessing the true potential of Human-AI Collaboration and achieving sustainable improvements in efficiency and Employee Engagement.

While the technological landscape is rapidly changing, the reliance on people is expected to remain a constant. The vision is not one where humans are entirely replaced, but rather augmented by AI, allowing them to focus on more meaningful and complex work. This Human-Centered AI perspective emphasizes that technology should empower individuals and enhance their capabilities, a crucial tenet for navigating the future of work with Trust and confidence.

Conclusion: From Insight to Action with Human-Centered AI

This exploration with Tuhin Chakraborty of Mimica AI has illuminated the path from deciphering complex workflows through AI-powered task mining to envisioning a future of agentic automation. By prioritizing data-driven insights, fostering Human-Centered AI with robust privacy measures, and preparing for a collaborative human-AI workforce, leaders can demystify new technologies. The journey to Digital Transformation is about understanding current processes deeply, identifying true opportunities for Automation, and empowering your teams to embrace the future of work, turning AI hype into measurable, human-centric success.

Ready to Transform Your Processes?

Curious about how these insights could reshape your organization’s efficiency? Discovering where Automation can truly make an impact begins with a clear understanding of your current processes. The team at Mimica AI offers a proof of concept to help enterprises embark on this journey, analyzing existing workflows to unlock a practical roadmap for transformation. You can explore how to get started by visiting mimica.ai/contact.

For those looking to delve deeper into leveraging data for strategic advantage, Tuhin and his team also provide an insightful resource: the “CxO’s Guide to Data-Driven Cost Cutting.” This white paper showcases how leading Fortune 500 companies are utilizing real data from their business processes to identify significant cost-cutting opportunities and effectively measure the ROI of process improvements. You can download your complimentary guide here: https://www.mimica.ai/white-papers/the-cxos-guide-to-data-driven-cost-cutting.

Join the Conversation

As we navigate the evolving landscape of Human-AI Collaboration, how is your organization identifying and prioritizing tasks for AI-driven Automation, and what has been the most surprising insight from your journey so far? Share your experiences and thoughts with the Innovation Tales community on our LinkedIn page: https://www.linkedin.com/company/innovation-tales.