Looking for a way to quickly and effectively incorporate GenAI tools and solutions into your business workflows to boost profitability? Learn more.

Blog

Generative artificial intelligence (GenAI) is more than just a buzzword in today’s corporate lexicon. Since the introduction of OpenAI’s large language model-based ChatGPT 3.5 in Nov 2022, it has become a catalyst for the transformation of traditional workflows. Its introduction has fundamentally reshaped various business functions, customer interactions, financial services, long-term strategy, investment planning, professional services, and talent development in ways that may have seemed impossible a few years ago. GenAI’s ChatGPT 3.5  broke the record for technology adoption, gaining 100 million users in the two months following its release.

It’s worthwhile to remember that this technology inflection point occurred only two years ago. GenAI’s rate of adoption and its impact have far outpaced those of any other business technology in recent memory. The release of the open source Deepseek-V3 model in Jan 2025 is another major milestone in the GenAI world, notable due to Deepseek’s significantly reduced cost of development relative to that of its predecessors: $5.6 million compared to the $78 million it cost to develop ChatGPT and the $191 million required to develop Google’s Gemini Ultra. Deepseek is also notable thanks to its developers’ claims that only 2000 Nvidia H800 chips were required to train its LLM.¹ Given these recent advances, there is no reason to believe the pace of change will slow anytime soon. While today’s GenAI models may ultimately be eclipsed by subsequent high-performing large foundational models and platforms, the GenAI revolution they have catalyzed will not be reversed.²

In 2025, large enterprises are expected to prioritize strategy, add business-IT partnerships to assist with GenAI projects, use GenAI models to take over many repetitive tasks, and move LLM pilots to production. According to a recent report by Forrester, by 2025, 750 million apps will use LLMs, underscoring the GenAI market’s rapid growth. Forrester predicts the market will grow in value from $1.59 billion in 2023 to $259.8 billion by 2030.³ 

Unlike much capital expenditure (CapEx) and technology spending, which often takes several years to see a return on investment (ROI), incorporating GenAI in organizational practice—even at the early stages of an organization’s implementation journey—can often increase earnings before interest, taxes, depreciation, and amortization (EBITDA), a measure used to evaluate a business’s core profitability.¹ The near-term productivity boost generated by the use of GenAI can be compelling and, over time, these initiatives may birth strategic investment opportunities to create valuable revenue generating assets.

While the benefits of size and scale once provided decisive advantages for larger firms in terms of access to specialized talent and capital-intensive infrastructure, the evolution of GenAI as a technology—particularly the development of GenAI as a service, the emergence of streamlined platforms, and the growth of customizable models—is leveling the playing field for mid-sized firms. 

The use of GenAI is not limited to technology companies alone. Healthcare companies such as Novartis and Moderna use ChatGPT Enterprise or Microsoft Copilot in their daily drug discovery and development research-based workflow processes. They have deployed GenAI enterprise tools to thousands of employees, not limiting the use of new technologies to only the upper echelons of leadership and ivory tower technologists. Democratized AI empowers all employees by creating novel use cases and expanding the impact of every team. 

Let us now look at the constituents of an LLM. Familiarity with these terms will suffice at this stage. We will dive deeper into these concepts in future editions of the blog. 

• Data represents the content used to train the model. LLMs typically must be trained on anything from terabytes to petabytes of data.
• Architecture is an attribute of the model itself, such as the number of parameters, content size, model embedding, tokens or size of the model. 
• Training models are further trained specifically to the desired use case, including chat, completions, or instruction. 
• Finally, fine-tuning is a feature added to models that refines the input data and model training to better match a particular use case or domain data set.

Fig. 1: LLM constituents

How can employees with varied skill sets at all levels of an organization upskill, adopt new technologies, such as LLMs, and successfully navigate the integration of these technologies to increase the productivity of their teams, improve the efficiency of business processes, and help foster the organization’s revenue growth? Given the technology’s newness and rapid evolution, many organizational leaders are unsure which tools to select, which use cases to start with, and how employees can make the most of GenAI opportunities. 

C-suite executives looking to harness the power of GenAI should distinguish between two distinct types of AI implementations: GenAI tools and GenAI solutions.

• GenAI tools enable their workforce to enhance individual productivity across myriad functions.⁴ GenAI tools include conversational AI systems, such as OpenAI’s ChatGPT, and digital assistants embedded in existing productivity software, such as Microsoft Copilot and Adobe’s Acrobat AI Assistant.
  

Moderna recently required all employees to be trained in GenAI tools, believing that the use of broadly applicable AI tools is a fundamental skill necessary for all job functions, despite the fact that drug discovery and development—not technology—is the company’s core mission.⁵

• GenAI solutions are business case-driven development initiatives that address strategic objectives, chosen based on innovation criteria for end-user desirability, technical feasibility, and business viability. These solutions are designed for specific teams and stakeholders that will generate financial returns directly via customer-facing products or indirectly via changes to processes, systems, and offerings at scale.⁴ 

Irrespective of which implementation pattern they choose, there are two approaches for deploying GenAI tools and solutions in organizational products, business practices, and workflows: Buy and Build-and-Boost. Buy and Build-and-Boost tools helps dissuade groups of stakeholders from independently pursuing unsanctioned GenAI tools and solutions when employees’ growing interest in new GenAI solutions is not addressed.⁶

Buy: One option is leveraging the out-of-the-box, no-code solutions available from GenAI providers and rolling them out to non-technical staff and managers by making them citizen developers of GenAI-infused automation apps, since they have the domain expertise to envision and develop these solutions. For example, Novartis and Moderna HR professionals use ChatGPT Enterprise for tasks such as writing job descriptions for new roles or drafting communications about policy changes, using GenAI as a starting point rather than beginning from scratch. Both companies provide access to the tool to all employees across their organizations, offering the freedom to innovate across all company functions, from clinical trials to corporate branding. 

The “Buy” approach is a good one for companies to adopt early in the GenAI journey, enabling them to explore and experiment with off-the-shelf GenAI solutions that are often opaque and geared to a narrow function in an industry vertical. Trying a few GenAI tools from trusted vendors, supported by hands-on training for citizen developers, with close oversight to manage risk and costs, is a safe, efficient way for companies to get started with GenAI.⁷ 

Build-and-boost: Another option is rolling out infrastructure as a service (IAAS) for technically savvy, hands-on-keyboard engineering staff.  For instance, GenAI providers such as Microsoft Corporation, Google, and Meta have significantly reduced the need for up-front investment and extensive IT capabilities by offering models and infrastructure as a service. Streamlined platform solutions allow companies to deliver business outcomes in financial performance (revenue growth, profit margins, and cost reductions), operational efficiency (improvements in productivity and process quality), customer engagement (satisfaction and retention), data quality enhancement, and talent development. At Moderna, more than 750 custom GPTs have been created across functions, with each user averaging 120 ChatGPT Enterprise conversations per week.⁷

With the build-and-boost approach, organizations can enhance vendor-provided models with their own proprietary data to fine-tune performance for specific contexts, offering more tailored results. This approach allows organizations to improve and customize, move fast, and build, to create a differentiated, competitive advantage. Common model fine-tuning and boosting approaches include the employment of either open-source large foundational models (LFMs) or proprietary LFMs in conjunction with Retrieval Augmented Generation (RAG) architecture that appends contextual domain-specific propriety data to improve model accuracy and provide customized solutions. We will cover the technology concepts and steps involved in the build-and-boost approach and RAG architecture in future editions of this blog series.

Whether you choose the buy approach or the build-and-boost approach, having a unified approach to determining which GenAI tools hold the most potential for your organization can reduce technology sprawl. Maintaining this unified approach will help you balance quarterly deliverables without compromising your long-term technological road map and strategic goals. Selecting the right GenAI tools for your organization means carefully navigating trade-offs in transparency, context-awareness, and cost.  Ensuring that your approach is cohesive will obviate many challenges, such as the ballooning cost of licensing for multiple technology platforms with similar functions using multiple vendors once free trials and early adoption incentives expire—a problem often faced in siloed organizations where decision-making around GenAI tools is decentralized and ad hoc. 

Embracing GenAI in business means being open to radical change, questioning existing business processes without fear of disrupting the status quo, and being dauntless in throwing out the rulebook and starting anew to achieve better business outcomes. Trailblazers, innovators, and those who are curious and on the lookout for technological developments that lie around the corner will reap the greatest benefit from GenAI. AI will not replace the role of humans in critical functions, but those incapable of embracing AI technologies will find themselves at a disadvantage, unable to partner and collaborate with AI practitioners within their organizations and beyond.

References:

1. https://source.washu.edu/2025/02/washu-expert-how-deepseek-changes-the-ai-industry/
2. https://hbr.org/2023/11/genai-could-transform-how-health-care-works
3. https://www.forrester.com/blogs/spend-on-generative-ai-will-grow-36-annually-to-2030/
4. https://cisr.mit.edu/publication/2024_0901_GenAI_VanderMeulenWixom?utm_source=pressrelease&utm_medium=pr&utm_campaign=twofacesofai
5. https://hbr.org/2023/09/where-should-your-company-start-with-genai
6. https://mitsloan.mit.edu/press/how-to-manage-two-faces-genai-tools-and-solutions-business-success
7. https://openai.com/index/moderna/