To effectively augment knowledge workers, businesses must artfully combine, orchestrate, and engineer multiple GenAI and AI models.
Enterprises have made major strides in transforming work processes. Automation efficiencies and productivity gains have been realized through AI and machine learning, analytics, and various software solutions. With the emergence of GenAI models, a vast frontier of opportunity also awaits to boost the effectiveness of knowledge workers.
Due to its highly nuanced nature, knowledge work, such as underwriting for banking or developing clinical protocols for drug trials, has been difficult to address through traditional AI. Because knowledge work typically involves intricate problem-solving with no clear-cut answers, decision-making is evaluated based on quality.
Engineering AI augmentation capabilities for improved enterprise decisioning at scale is no trivial matter. A new choice architecture is needed to build AI systems that can deliver domain reasoning reliably and safely.
Moreover, no single AI model exists that can approximate how a human brain thinks and reasons. Building an ultimate model is an impractical goal. There is no universal definition for knowledge work which tends to vary widely among industries, enterprises, and business functions.
The ultimate challenge is posed by trying to replicate the complex ability of critical reasoning in knowledge work for making effective decisions based on different contexts with consistent results. Today, it is possible for industry leaders to address this challenge for business functions with high knowledge concentration. The constant nudging of a formal digitalized and seamlessly integrated AI augmentation capability can provide an organization’s knowledge workers expert guidance on the right decision to make in different contexts.
In other words, “Knowledge work transformation is now possible to achieve as an output of knowledge worker transformation through personalized AI augmentation,” explains Sankaranarayanan “Shanky” Viswanathan, Vice President and Head of Strategic Business Initiatives, Corporate Technology Office, TCS. “Engineering AI augmentation capabilities for improved enterprise decisioning at scale is no trivial matter. A new choice architecture is needed to build AI systems that can deliver domain reasoning reliably and safely.”
Enabling a personalized AI augmentation capability will boost the overall performance of knowledge workers while lowering the skill threshold.
Enterprise GenAI tools can be used to discern and encode the tacit knowledge and reasoning ability of an organization’s subject matter experts as a digital capability. A system of models and tools must be carefully orchestrated and built to enable this capability. An AI-powered genius partner, or specialist copilot, can then effectively augment other knowledge workers across the enterprise with that expert knowledge. Democratizing knowledge and empowering employees to perform at an expert level will yield constant, consistent, and predictable business outcomes.
Moreover, leveraging such expertise at scale is not just about helping employees make the right calls when decisions are needed—it is equally about not making the wrong calls which could have implications for the enterprise. In essence, businesses can advance their use of AI from enabling automation efficiencies for noncore, repetitive work tasks to boosting the overall quality of core, knowledge-intensive tasks that deliver higher value.
As options for AI and GenAI models continue to grow, so does the business community’s optimism for applying them. The Working Towards the Future Report prepared by TCS, based on a global survey of prominent futurists, executives, and foresight professionals, concluded that 90% of them are optimistic about AI-driven changes to how people work. AI is expected to level the playing field by providing workers of different levels equal access to information and capabilities, driving increased engagement, ideation, and innovation.
Based on the TCS AI for Business Study Key Findings Report conducted globally, CEOs and other senior execs surveyed believe the impact of AI will be greater or equal to that of the internet (54%) and smartphones (59%). However, relatively few of the respondents (only 4%) have leveraged AI in a way that is transformative for their business.
An integrated, AI-first approach can facilitate the building of a reasoning system based on the right augmentation strategy for addressing enterprise-specific intelligence needs.
What does it take to build an advanced reasoning system? First, it’s important to understand that intelligent AI reasoning used in complex decision-making generally involves multiple methods, such as:
For instance, a contact center may primarily leverage logical or analogical methods. Augmenting clinical trial scientists, insurance underwriters, or retail category managers, however, will typically depend on more complex experiential reasoning.
Determining the right AI models to mimic human reasoning is therefore as much an art as it is a science. While traditional AI might be the sensible choice for some predictive intelligence scenarios, the type or lack of data might require a GenAI model. Or if a particular scenario demands more preciseness, a GenAI model may need fine-tuning for optimization. In other instances, some activities that are more structured in nature with access to standardized data may not necessitate a GenAI capability at all.
Enabling a robust reasoning system that can effectively perform myriad intelligence functions is essential for implementing a personalized knowledge augmentation strategy. A composite approach can facilitate the need to toggle between different models in addressing the specific intelligence needs of a business. Rather than trying to model the multifaceted functions of an entire human brain, the focus is on understanding and modeling how an individual performs a very narrow or finite set of knowledge work activities.
Augmenting the decision-making abilities of knowledge workers in selectively targeted areas will yield elite outcomes with higher returns.
There are many opportunities to drive value creation through the personalized augmentation of workers in knowledge-intensive business functions across industries. As CEOs, COOs, and business unit heads assess opportunities, they should focus on where they can increase, or potentially even double, their productivity without necessarily increasing their capacity (learn more about becoming a knowledge-driven enterprise).
A few standout examples of opportunities for reimagining knowledge work include:
To illustrate, consider the benefits of leveraging GenAI in a global architectural firm. Engineers and architects are tasked with balancing client design expectations against material performance and complex building codes and safety standards that differ between states and countries. They may also need to adapt designs and construction based on new materials or laws. But imagine if the firm could efficiently capture all relevant information learned on a project and distribute it for future use across the business. The shared experience and learnings could facilitate more effective decision-making and faster completion of projects.
The potential benefits of enabling personalized AI augmentation to advance the performance of knowledge workers are significant. To drive impactful transformation, however, businesses must step diligently through the journey. Selectively augmenting workers in strategically targeted areas of an enterprise versus taking a jackhammer approach is key to optimizing returns.
There are three levels of advancement to build and scale a reasoning system with a personalized AI augmentation capability.
For starters, modeling knowledge-intensive tasks for personalized augmentation involves working closely with subject matter experts to pinpoint where contextual experience heavily influences business outcomes. Driving success requires a deep understanding of the work's intricacies. From there, what does the journey to transform knowledge work actually look like? What are the key factors enterprises must consider along the way?
The transformation journey to build a reasoning system with an AI augmentation capability can be broken down into three major levels of increasing value and complexity:
Because enterprise data that exists in traditional form is engineered for processing and optimization by systems, it generally lacks the meaningful context required for processing by humans. One of the biggest misconceptions enterprises have is that they can simply feed such data to an LLM or other generative model. They expect that it will be able to automatically understand the data.
Generative models behave like human beings and must be fed data in a similar way. The models need enterprise context to understand the import of the pattern, including the insights and reasoning that will drive a particular outcome. In effect, it is important to understand that training a generative model is not akin to how traditional models get trained. Rather, a huge amount of effort, insight, and care is needed to curate, make available, and fine-tune the data for enterprise use.
The user interface of the reasoning system should be designed to facilitate a frictionless and immersive augmentation experience for the knowledge workers executing the business processes. The augmentation capability should be seamlessly woven into the existing systems of an enterprise to efficiently nudge workers along with the right directional inputs for making effective decisions.
Autonomous agents can significantly enhance AI augmentation capabilities while ensuring adherence to goals, rules, and constraints.
Today’s AI-driven models are not sentient, or self-aware, and developing such an ability is unlikely for the foreseeable future. They have the potential to get carried away by their creative abilities and fall out of alignment with the scope of work for achieving desired business outcomes. They may start to exhibit artificial insanity, acting erratically and generating nonsensical outputs. Without security guardrails in place to enforce enterprise policies, models may unknowingly share sensitive information and generate responses that are false, biased, or profane in nature.
Similar to how humans need to be trained on the expectations and rules of operation for an enterprise, generative models must also be trained on what is and is not permissible behavior. Otherwise, the models have no way to keep themselves in check and self-correct. That’s where the important role of autonomous AI agents comes into play (learn more about guardrails).
Everyone is especially concerned about GenAI’s propensity for hallucinations, but it’s actually an inherent feature and not a design flaw—just like humans, they are prone to overextend when their knowledge is insufficient to answer.
An agentic AI-based system must be built and enabled with specific engineering capabilities to train the models for enterprise use. AI agents must each be tasked with a specific purpose that aligns with the goal of a particular knowledge activity—from enhancing data analysis and generating domain reports to optimizing decision-making processes.
“Everyone is especially concerned about GenAI’s propensity for hallucinations, but it’s actually an inherent feature and not a design flaw—just like humans, they are prone to overextend when their knowledge is insufficient to answer,” explains Narendran Sivakumar, Global GTM Lead, Generative AI, Corporate Technology Office, TCS. “Hallucinations are core to how generative models creatively think and produce outputs. To make the models trustworthy for enterprise use, you need to build a highly structured reasoning system and optimize it with autonomous AI agents to effectively harness their creative genius.”
In a way, agents can infuse a nuanced form of sentience into the intelligence fabric by being purpose-driven and contextually relevant. They will continuously learn, self-correct, and iterate until the desired outcomes for a specified set of knowledge activities are achieved.
A multi-model approach to enterprise GenAI can be successfully enabled with an advanced reasoning system when underpinned by purpose-bound, autonomous AI agents. By empowering knowledge workers with a personalized AI augmentation capability, businesses can drive transformative business value across the enterprise and optimize their investment returns.