“There are only two types of companies in this world, those who are great at AI and everybody else. If you don’t know AI, you are going to fail, period, end of story. You have to understand it, because it will have a significant impact on every single thing that you do. There’s no avoiding it.” – Mark Cuban.
This quote sums up the need for companies to prioritize artificial intelligence (AI) initiatives and also captures the state of the AI race today. According to recent research by Boston Consulting Group:
- Only 4% of companies adopting AI have reaped significant value from the technology.
- 22% have implemented an AI strategy, built advanced capabilities and are beginning to realize substantial gains.
- 74% of companies are struggling to achieve value and scale.
Over the past three years, AI leaders in the 22% bracket have achieved 1.5 times higher revenue growth, 1.6 times greater shareholder returns, and 1.4 times higher returns on invested capital, according to the BCG survey.
AI is highly disruptive and rapidly evolving at breakneck speed. Advanced versions of large language models (LLMs) are coming out at regular intervals with improved compute power capabilities. Those LLMs are also achieving the kind of increased accuracy scores on the MMLU benchmark leaderboard that have become synonymous with the version upgrade cycles we are all accustomed to with SaaS product releases like clockwork.
Yes, you heard that right! That’s how much better the AI/ML models, tech stack, tools and products have gotten of late…and within an amazingly short time frame. In fact, China just unveiled “DeepSeek” with an advanced DeepSeek-R1 open-source, open-weight model that runs on a fraction of compute power used by ChatGPT, Anthropic and Gemini models. It’s already displayed stunning algorithmic efficiency and its outputs are on-par for different use cases. Some are already calling it a “Sputnik moment” for AI. It’s certainly sent shockwaves through the industry.
The real challenge for many companies, however, has been on how to crack this code (pun intended) that is (unsurprisingly) not accessible to everyone, and to ultimately generate that pot of gold that’s critical to many companies’ long-term future.
Strategic alignment
Companies are increasingly realizing that AI is not just a tool but a transformative force that can redefine business models, enhance customer experiences and drive operational efficiency. Developing a clear and comprehensive strategic vision is the starting point of prioritizing AI initiatives with business goals.
Answering the question “Why” sets the tone. Will AI — and specifically generative AI (genAI) — assist in your customer experience and retention efforts? Are you hoping it will open up or drive new lines of revenue? Or is it a “we’re not sure, but we don’t want to be left behind” situation? In a recent Gartner webinar poll of more than 2,500 executives, 38% indicated that customer experience and retention is currently the primary purpose of their genAI investments, while 26% say revenue growth as the focus of their AI initiatives.
Once your AI vision is in place, another key strategic alignment required here is to ensure AI initiatives are not pursued in isolation but integrated into the broader cohesive strategic framework of the organization. Pivot to a tech stack — preferably open source — that fits your IT landscape and does not require a major overhaul to retrofit. This coordinated approach should serve as a foundation principle to craft a good roadmap for AI initiatives, prioritizing use cases in business continuity or increased productivity categories, guiding their development, implementation and finally scaling the services.
Leadership and governance
Leadership is crucial in ensuring that AI initiatives align with business goals. CIOs and senior leaders must provide strong leadership, direction and, most importantly, supporting the roadmap. This is especially critical during the execution phase of the journey in order to navigate the bumps that are bound to arise at the business process, people and culture pillars. Effective leadership involves strategic planning and active engagement with AI initiatives, championing AI efforts, communicating their strategic importance, building a culture of alignment and ensuring they are integrated into the business strategy.
Most importantly investing in the right team of people to bring this vision to reality. For most of the 22% of AI leaders who have emerged, the time since ChatGPT went live in November of 2022 has been a period of AI incubation, investment, and focused effort. Much of that energy has been spent assembling the right team and fully backing them with robust leadership and change management. The dynamic nature of both AI and business environments warrants this approach because it requires continuous evaluation and adjustments of AI initiatives. That’s just the nature of this beast.
Governance starts with data and is then integrated into AI. Data-driven decision-making is at the heart of any successful AI implementation. Ensuring that data is used responsibly and compliantly is a prerequisite. As AI becomes more embedded in data processes, governance in AI encompasses data integrity and quality.
The governance framework must implement systems that validate and clean data, ensuring the AI models are built on accurate and reliable information. Databricks, Palantir and Snowflake are excellent product services that successfully serve this combined principle of data and AI for companies if they choose to go with commercial off-the-shelf (COTS) tech. Otherwise, the Big 3 cloud providers have niche tools available to companies if they choose to build a custom solution.
Metrics and evaluation
Establishing clear objectives and metrics for success ensures AI initiatives align with your business goals. A measurement framework helps ensure your company’s business goals are in line with the business strategy. Regularly assessing the progress of AI projects and their alignment with business goals is essential, as is being prepared to adjust or change direction based on feedback, performance metrics or changes in business priorities.
Executing AI initiatives through agile project management methodology is clearly helpful here. Key performance indicators (KPIs) and metrics can be categorized in two ways:
- Business-level metrics: These directly measure progress towards specific business goals.
- AI framework-level metrics: These provide an overarching assessment of the AI system itself. Examples include:
- Model quality: Accuracy, precision, recall, etc.
- Latency: How quickly the LLM generates output.
- Cost: Operational costs associated with running the AI system.
- Prompt engineering maturity: The effectiveness of techniques used to elicit desired responses from the LLM.
Systematic framework: A hypothetical case study
Say you’re working for a retail company to improve its supply chain. You have established the overall vision for AI adoption within the company. To establish the framework to maximize AI ROI, the following steps are necessary:
- Business use case: Leverage AI to improve supply chain operations (boost productivity and profits).
- AI strategy: Use the appropriate AI algorithms to analyze supply chain data and generate insights to optimize operations. This includes processing data, building models, sampling strategies and integrating feedback loops by leveraging appropriate AI tech stack (COTS or cloud provider niche LLMs). As part of the strategic alignment here, pick the right LLM tech stack that matches the company’s cohesive overarching purpose for all AI use cases.
- Business goals: Reduce lead times by 20 percent, increase on-time delivery to 98 percent and reduce transportation costs by 10 percent.
- Measurement framework: This is a set of metrics used to track progress. In this case, it includes lead times, on-time delivery, transportation costs, inventory turnover and order cycle time.
- Prioritize the AI initiatives: These include demand forecasting for better inventory management, route optimization to minimize transportation costs and warehouse automation to reduce labor costs.
The proof is in the pudding
The Maha Kumbh Mela 2025 Hindu pilgrimage underway at Prayagraj, Northern India, experiences the largest congregation of people visiting daily (an estimated 450–600 million people over 45 days). This year’s pilgrimage has been a true revelation of AI technology unraveling on a huge scale and value. AI technology with a coordinated framework has been implemented on a massive scale at this event and is helping the local governing authorities as well as the Indian government in such areas as crowd control, stampede prevention, logistics, transportation & parking, security and surveillance, information services, lost and found services of family members — all through AI chatbots, AI-enabled drones and AI/ML software solutions. The cohesively applied AI framework and tools are helping to tackle the logistic issues of this humongous scale successfully for the very first time in the history of this event. It’s a real-world example of technology meeting tradition!
Santhosh Gottigere is an IT Technology leader, who has led major business and IT transformations in enterprise architecture, service operations, human resources, coud migrations, and SAP ERP spanning across multiple vertical industries such as healthcare, pharmaceuticals, energy and gas, retail and manufacturing, consumer goods, and transportation and freight management across the US, Europe and Canada. He is a global information technology leader with broad experience from multinational IT consulting organizations and leading independent software vendors.
This article was made possible by our partnership with the IASA Chief Architect Forum. The CAF’s purpose is to test, challenge and support the art and science of Business Technology Architecture and its evolution over time as well as grow the influence and leadership of chief architects both inside and outside the profession. The CAF is a leadership community of the IASA, the leading non-profit professional association for business technology architects.