How to maximize on the AI Investments?
Take the right first steps and a strategic approach towards successful implementation of AI Projects.
GenerativeAI (GenAI) has created a new wave in technology since November 2022.
GenAI stands at the forefront of the current technology transformation wave for all the right reasons.
It can generate content, augment knowledge, and use large language models (LLMs) to respond to any prompt previously achieved by human creativity.
In the last two years, GenAI has gone through multiple Explorations. By exploration, I mean it has gone through various hypothesis tests - understanding the product-market fit and the value it delivers.
According to Harvard Business Review, GenAI has reached the peak of demand or Extrapolation as its next phase of Venture Growth.
This surge in demand for GenAI signifies a strategic shift in the business landscape. Companies now focus on generating additional revenue at optimal costs to achieve a profit-market fit.
So, for companies to get the most out of GenAI, what are the key focus areas and best practices that will help them maximize the ROI to achieve competitive advantages
Well Defined Strategy
In the current exploration phase, companies are beyond the experimenting stage. Early adopters are now investing in building GenAI tools or embedding GenAI in existing tools.
The key business question about AI, in general, is whether it should be a product or a feature. Opinions differ based on the use cases.
With a lot of controlled experimentation, AI-leading companies are taking both approaches.
There are a lot of GenAI tools promising improved quality of LLM response and making it relevant to the context of the prompt.
On the other hand, existing tools in well-known ecosystems like Microsoft, Google, and Apple - GenAI is embedded as a feature to boost productivity.
Companies implementing GenAI for their enterprises must have a defined strategy for better adoption. Understanding the key objectives helps achieve success and drives and measures the business value generated by GenAI implementations.
Following are key business questions that will help with defining strategic objectives for GenAI projects -
What is the scope of the GenAI project, and how is it measured?
Is the scope to boost productivity at the individual level, OR is the scope very complex, with multiple changes resulting in the acquisition of new technologies?
What is a capital investment into GenAI projects, and how is the change propagated through the organization?
For maximum benefit from GenAI projects, there needs to be a commitment of capital & human investment.
Such projects are paradigm shifts to a new norm that require behavioral changes, such as adopting and learning new skills and technologies or updating and changing business processes to align with GenAI or machine interactions and interpretations to deal with Hallucination (flawed interpretation).
For driving GenAI adoption with the clear intent of shifting such technology to augment human capabilities rather than replace or reduce human interactions through automation.
What is the business value delivered by the GenAI project, and is it measurable?
Organizations have good documentation of business values; however, the quantification of business value should be more precise and agreed upon during the preliminary phases.
According to Harvard Business Review, law firms like A&O Shearman and Wilson Sonsini have invested in AI tools for contract review. So, KPIs like - GenAI tools make the contract review process faster, more efficient, and optimized with fewer errors, which are measurable business benefits of GenAI.
Similarly, Sanofi uses GenAI and other AI tools to optimize the time to market for new drugs. These benefits result in additional revenues and increased profit for the organization.
A Mature Data Management Practice
In the last several years, organizations have not only dealt with structured data but also accumulated unstructured data in videos, text, images, etc.
Various research studies show that organizations face challenges in unifying data from multiple sources and fueling analytics to drive value and make critical business decisions based on data.
Organizations adopting GenAI must have well-defined processes for collecting, storing, and tagging trusted data sources.
But how do organizations mature to a state where data is understood well by all stakeholders?
Following are key decisions for the successful implementation of the GenAI strategy -
Build VS Buy:
Organizations must choose whether to build a new GenAI platform or invest in a third-party vendor platform to augment the existing capabilities and data for faster ROI.
GenAI platform will learn from the enterprise data sources to evaluate and assess the critical data elements, their uniqueness, and their accuracy in building an enterprise-owned representation of data (as well as statistical models and large language models -Enterprise LLMs) that are well-understood and trusted by all stakeholders.
Such a platform can integrate with publically available LLMs for a broader context. In such cases, the security, privacy, and reliability of sources that generate content must be validated and governed through human interactions.
Building such a platform may not be optimal for TCO. Some companies prefer investing in third-party vendor capabilities. However, the challenge would be bridging the gap between enterprise data and how such a platform treats it. Sometimes, building a semantic layer (in the form of Virtualization or Data Lakehouse) helps.
Choice of practice use cases for quick wins:
With technology projects, organizations often take longer to realize value and drive adoption for new initiatives. Most of the time, the initiative fails to deliver due to the least viable use cases. The following are essential use cases for initiating any GenAI project based on the maturity of an organization -
Improving data management practices is essential in adopting AI. According to a survey conducted by Microsoft, organizations need help to integrate diverse data sources into a unified form. Other challenges lead to untrusted data within an organization that could be improved.
Establishing data governance policies and frameworks, consistently managing the end-to-end data lifecycle through data quality(DQ tools) and master data management (MDM) tools, and cataloging data lineage are critical use cases for mastering and managing data efficiently.
Breaking data silos by using tools like data virtualization for consolidating data into a data lakehouse for storing and analyzing data in any form (structured, unstructured) will help with unifying data from various sources. It will also help with uniform understanding of data across all stakeholders.
Advanced technologies like graphs and vector databases will help establish relationships and faster data analysis. The data, which included text, images, audio, and video, is analyzed using vector databases to perform semantic and sentiment analysis, further demonstrating the value proposition of AI investment.
Last but not the least, you need “Real Data” for fine-tuning your models. The question is how to generate real-data to test the AI Models for optimal output. For more details, please read-more about hands-on examples of generating data for AI Projects
Key Takeaways -
Similar to previous technology waves like Big Data and Cloud migration, AI has renewed the urgency of unifying data and improving quality to leverage the strategic value of data through GenAI capabilities for boosting productivity and achieving success through additional revenue from AI-based business automation.
So, the question is whether or not your organization is getting serious about these focus areas to drive business value through GenAI Investments.