Erik Schwartz is the Chief AI Officer (CAIO) Tricon Infotech. a leading consulting and software services company. Tricon Infotech delivers efficient, automated solutions and full digital transformations through custom products and enterprise implementations
Erik Schwartz is a seasoned technology executive and entrepreneur with over two decades of experience in the tech sector, specializing at the intersection of AI, Information Retrieval and Knowledge Discovery. Over the course of his career, Erik has been at the forefront of integrating building large-scale platforms and integrating AI into search technologies, significantly enhancing user interaction and information accessibility. His previous held key senior roles at Comcast, Elsevier, and Microsoft, where he led pioneering AI, search, and LLM initiatives.
Erik’s professional journey is marked by his dedication to innovation and his belief in the power of collaboration. He has consistently driven teams towards the swift delivery of groundbreaking solutions, firmly establishing himself as a trusted leader in the technology community. His work, most recently on the Scopus AI project at Elsevier, underscores his commitment to redefining the boundaries of how we engage with information and create a trusted relationship with users.
In his role as Chief AI Officer (CAIO), Erik leverages his extensive experience to develop and implement comprehensive AI strategies for Tricon customers. His thorough process not only demystifies AI but also ensures that these businesses are equipped to succeed and thrive in the competitive landscape of AI technology. Erik is passionate about fostering growth and innovation, sharing his insights to inspire and empower organizations to harness the transformative power of AI effectively.
Can you share some highlights of your career journey that led to your current role as Chief AI Officer at Tricon Infotech?
I have been immersed in the Information Retrieval domain throughout my entire career. My journey began in the early 90s as a Web Master at the dawn of the Internet. During this formative period, I focused on building digital libraries for government agencies, universities, and media companies, which laid the foundation for my expertise in digital information systems.
In the 2000s, I transitioned to working with Search Engine vendors, where I honed my skills in search technologies. This phase of my career was marked by significant growth and learning through various acquisitions, ultimately leading me to join Microsoft in 2008. At Microsoft, I played a pivotal role in developing and enhancing Knowledge Discovery Platforms, driving innovation and improving information accessibility for users.
Following my tenure at Microsoft, I led initiatives at major corporations such as Comcast and Elsevier, where I was responsible for running large-scale Knowledge Discovery Platforms. These experiences have been instrumental in shaping my approach to AI and information retrieval, culminating in my current role as Chief AI Officer at Tricon Infotech. Here, I leverage my extensive experience to drive AI strategies and solutions that empower our clients to harness the full potential of their data.
How have your experiences at companies like Comcast, Elsevier, and Microsoft influenced your approach to integrating AI and search technologies?
Throughout my career, I have been deeply focused on natural language processing (NLP) techniques and machine learning. Initially, these technologies were based on simplistic rules-based systems. However, as data sets grew larger and computing power became more robust, we began to significantly enhance user experiences by automatically harvesting data and feeding it back into the algorithms to improve their performance.
At Microsoft, following the acquisition of FAST, I served as a product manager on the SharePoint team. In this role, I was involved in integrating advanced search technologies into enterprise content management systems, enhancing information retrieval and collaboration capabilities for businesses.
At Comcast, I built a knowledge discovery platform that powered their entire video business, enabling users to search and discover content across set-top boxes, mobile, and web devices. This search engine scaled to handle over 1 billion requests per day, significantly improving the user experience by providing fast and accurate content recommendations and search results.
One of the most transformative experiences was at Elsevier, where we launched a Generative AI experience for Scopus, one of their most trusted products. This initiative utilized a Large Language Model (LLM) to assist users in asking better questions and obtaining more accurate answers from the deeply technical content in the scholarly communications database. This LLM-driven approach ensured the complete accuracy and trustworthiness of over 90 million articles contained within the database, demonstrating the power of AI to enhance academic research and knowledge dissemination.
What excites you the most about the current advancements in Generative AI and its potential applications?
One of the biggest historical challenges in Information Retrieval has been maintaining context. For humans, this is a natural process, but for machines, finding information has traditionally been a very transactional experience: ask a question, get an answer. Diving deeper into a topic required asking increasingly specific questions. Generative AI revolutionizes this approach by enabling a more conversational and contextual interaction, much like a natural conversation with someone you’ve just met.
Furthermore, Generative AI incorporates additional techniques that enhance deeper understanding, which have historically been difficult for traditional search engines. For example, Large Language Models (LLMs) can seamlessly handle aspects such as tone, sentiment analysis, semantic understanding, and disambiguation. These capabilities allow LLMs to grasp the nuances of human language and context effortlessly, providing more accurate and meaningful responses right out of the box. This advancement excites me the most, as it opens up a myriad of possibilities for creating more intuitive, responsive, and intelligent applications across various domains.
How does Tricon Infotech’s approach to GenAI differ from other companies in the industry?
In the Generative AI space, there are two primary focus areas. The first, which receives significant attention from some of the largest technology vendors, is training and fine-tuning AI models. The second area, where Generative AI practitioners truly excel, is inference—using Generative AI to create valuable products and services.
At Tricon Infotech, we focus on the latter. Our approach is distinct because we emphasize practical application and rapid deployment. We have developed a comprehensive program that helps business leaders quickly identify the most impactful use cases for Generative AI. Our process includes a rapid prototyping solution, enabling customers to work with their own data in an AI sandbox. This approach ensures that they can see tangible results and engage with AI-driven insights early in the development cycle.
Moreover, we have a radical focus on time-to-value. Our goal is to help customers build and deploy consumer-facing applications within 90 days. This accelerated timeline not only drives faster innovation but also ensures that businesses can quickly capitalize on the benefits of Generative AI, creating new revenue streams and enhancing customer satisfaction.
Can you discuss some of the key challenges in implementing Large Language Models (LLMs) and Generative AI in enterprise solutions?
Implementing Large Language Models (LLMs) and Generative AI in enterprise solutions presents several emerging challenges. The first and foremost challenge is trust. Enterprises must be assured that AI systems will not compromise their intellectual property or sensitive corporate information. Ensuring data security and obtaining proper assurances that the AI will not misuse data is critical for gaining trust.
The second challenge is the issue of hallucinations. Generative AI can sometimes produce confident answers that are factually inaccurate. This can undermine the reliability of AI systems. Techniques such as fine-tuning models and employing Retrieval Augmented Generation (RAG) can help mitigate the occurrence of hallucinations by ensuring that AI responses are grounded in accurate data.
The third significant challenge is cost. The licensing and scaling of LLMs can be quite expensive. Even enterprise offerings from major providers like Microsoft, Amazon, and Google come with steep entry fees and minimums. Therefore, it is crucial for enterprises to closely monitor and manage the return on investment (ROI) to ensure that the deployment of AI solutions is economically viable.
Can you explain the structured approach Tricon Infotech uses to develop customized GenAI enterprise solutions?
Tricon Infotech is a product development company that stands apart by offering managed services through dedicated, full-stack product teams rather than traditional staff augmentation. Our approach involves deploying entire product teams that can manage every aspect of the product development lifecycle, including user research, user experience design (UX), front-end and back-end development, test automation, deployment, scaling, and ongoing operations.
This comprehensive managed service model ensures that our customers can focus directly on capturing value from their data without the complexities and overhead of managing separate resources. Our key driver is time to value, meaning we prioritize delivering tangible benefits quickly and efficiently. Our ambition is to build long-term generative relationships with our customers by continually adding value and iterating through the feature development process.
Our structured approach is designed to be agile and responsive, enabling us to adapt quickly to new challenges and opportunities in the AI landscape. By leveraging the full capabilities of our multidisciplinary teams, we deliver highly customized Generative AI solutions that are tailored to the specific needs of each enterprise. This approach not only differentiates us from traditional staff augmentation firms but also ensures that we provide holistic, end-to-end solutions that drive significant business impact.
What are some examples of real-world problems that Tricon’s GenAI solutions have successfully addressed?
- E-Learning – converting traditional media and legacy educational material into interactive multi-modal content. This allows our customers to repurpose existing content to adapt to new ways of learning and reach learners on different platforms where they already are. Further, the content can then be repurposed into hyper-personalized learning programs that can adapt automatically to the learner’s needs and learning styles (audio, visual, etc.)
- Private AI – Helping customers build trust enterprise AI solutions that remain private and honor customers access rule, while maintaining costs and helping to scale out across the various functions of the enterprise helping overburdened professionals and shared services scale out better to the organization while natively understanding the various rules and restrictions of locale and regional policies distributed geographically. These private Ais will not only serve the enterprise but will also generate new streams of revenue for our customers.
- Process Automation – there are still a massive number of organizations who rely on manual processes and swivel chair data integration. AI helps to connect the various system together by creating intelligent layers that not only can validate data, but can understand the bespoke signal created by the unique dataset or tooling and help efficiently route workflows around while identifying supply chain issues
What role does continuous learning and growth play in staying ahead in the rapidly evolving field of AI?
One of the most significant challenges in the AI field is upskilling the talent pool. There is a new generation of workers who intuitively understand AI tools and technologies. However, there is also an older generation that needs to grasp what these tools can and cannot do. Continuous learning is crucial for bridging this gap.
AI tools have the potential to dramatically enhance productivity, allowing businesses to achieve much more with significantly fewer resources, thereby reducing timeframes and costs. For these benefits to be realized, employees must be open to learning new ways of working and integrating these tools into their workflows.
Moreover, addressing the fear of job security is essential. Employees must understand that those who embrace continuous learning and growth will be better equipped to incorporate new AI tools into their daily routines, ultimately leading to greater job security. The reality is that success in the AI-driven future will come to those who actively seek to understand and leverage these evolving technologies.
How do you envision the future of AI transforming search technology and user interaction in the next decade?
We are already witnessing a significant shift from traditional search engines to Generative AI tools for initial queries. This shift is driven by the ability of Generative AI to provide direct answers and solutions, eliminating the need to traverse multiple websites or resources independently. In the near future, it will become commonplace for AIs to attend meetings, take actions, and handle routine tasks, leading to a substantial reduction in the roles of certain functions within enterprises.
One of the key challenges that remains is figuring out how to monetize Generative AI, as the traditional advertising model may face significant hurdles in this new landscape. My prediction is that data will become increasingly valuable, acting more like a currency as we navigate this brave new world. This shift will require innovative business models that leverage the unique capabilities of AI while ensuring that users and enterprises can derive tangible value from their interactions.
Overall, the future of AI in search technology and user interaction promises to be transformative, making information retrieval more intuitive and efficient while reshaping the way we approach digital interactions and enterprise functions.
What practical advice would you give to businesses looking to leverage AI to drive success and innovation?
Don’t be afraid of the technology. Start by making AI tools available to your employees to ensure that your data and intellectual property (IP) remain secure. Many employees are already using AI tools, but without proper governance, there is a risk of misuse. Therefore, it is crucial to upskill your staff so they understand the risks involved and how to use these tools safely and effectively.
Additionally, it is essential to pay close attention to the measures of success. AI tools can be expensive, but the costs are expected to decrease over time. However, it is important to keep a clear focus on the return on investment (ROI) to manage costs and understand the impact on your business. By doing so, you can leverage AI to drive innovation and success while ensuring that the benefits outweigh the expenses.
Thank you for the great interview, readers who wish to learn more should visit Tricon Infotech.
