Syniti CEO dishes out generative AI playbook for enterprises

Kevin Campbell, Chief Executive Officer, Syniti. Image courtesy of Syniti.

Generative AI has undoubtedly entered the consciousness of enterprises, yet many questions and challenges remain, particularly in data governance. How can organisations gain an early advantage in the generative AI race and explore new ways to enhance their business?

Syniti, an enterprise data management provider, is one such company harnessing the power of generative AI to increase its operational agility. Kevin Campbell, the company’s Chief Executive Officer, addresses several pressing questions about the technology and details how Syniti is benefiting from its capabilities.

What are the most critical data attributes that businesses should focus on to ensure the reliability of their generative AI outputs?

Firstly, it is important to understand the objective of using generative AI and define the outcomes that you’re hoping to gain. For example, solving specific business problems or driving efficiencies in operations. This will help businesses understand how they are going to train the models and examine the underlying foundation of data needed to achieve these outcomes.

To get the results that are aligned to these expected business outcomes, there are two critical components: choosing the right data and the quality of that data. These two are closely interlinked.

Data quality is critical; the adage “garbage in, garbage out” strongly applies here. If you have the right data sets, but the quality is poor, it may result in issues such as inaccurate recommendations, result bias, and irrelevant guidance. Similarly, if you have quality data, but it’s the wrong data — meaning, data is not of the right context — the result may not serve the intended business outcomes at all. It’s a matter of bringing together the proper curation of data and high data quality to ensure that you can leverage the power of generative AI.

To that end, some of the critical data attributes include:

  • Relevance, accuracy, and completeness: This is the foundation of “good data” and reflects the common business term, “single source of truth.” It is important to ensure that the training data is representative of actual business scenarios, accurate to the tee, and free from errors. This means, no duplicated, outdated, or poorly formatted data. The data must be consistent, regularly updated, and in-context to the outcomes you are aiming to achieve.
  • Diversity: To prevent bias, it can be helpful to include a diverse set of examples or scenarios to help the model produce outcomes for different situations.
  • Compliance and privacy: It is crucial to ensure that the data used in AI models adheres to privacy regulations and standards, such as GDPR or HIPAA. Furthermore, to protect individuals’ privacy, anonymise sensitive information.

These are foundational data attributes that must be nailed early on. Thereafter, businesses must focus on training the AI engine to produce explainable and interpretable results, and subsequently, refine the results through review and validation to improve speed and accuracy of the final outputs.

In the context of generative AI, how do you foresee the role of data governance evolving in the next five years?

Data governance will definitely play a more central role in business’ data strategies. Here are some of the trends that will bring data governance to the forefront:

  • AI governance and ethics: In recent years, we’ve seen government bodies and agencies introduce AI governance frameworks, and we can expect more updates in the future, with a focus on ethical considerations. For instance, in May 2022, Singapore’s Infocomm Media Development Authority introduced an AI governance testing framework for companies that wish to demonstrate responsible AI. More recently, ASEAN members gathered to discuss the development of a guide on AI governance and ethics to balance the economic benefits of the technology with its risks. These guidelines will need to be incorporated in data governance frameworks.
  • Data quality: Data governance practices will be crucial in the training stage, and ensure that businesses are able to retain the data attributes previously mentioned to achieve its intended business outcomes.
  • Decision-making: In higher-stake applications of the technology, such as healthcare or banking, understanding why an AI model has made a specific decision is important for users to accept the results and build trust in the technology. Furthermore, it is important for non-technical users to be able to understand the model’s logic to trust the decisions being made. These elements of explainability and interpretability necessitate the need for data governance frameworks as it manages all aspects of data, from data quality management and user feedback integration to ensuring compliance.
  • Regulatory compliance: Regulatory frameworks are constantly evolving, and new ones may emerge. Data governance will be key to quickly adjust to these changes that businesses must comply with.

In what ways is generative AI transforming Syniti’s internal processes?

Our focus with generative AI is to make our people more productive with what they do every day — whether that is storing and cataloguing content reuse more efficiently, better answering product support questions, responding to client sales opportunities, or finding the most relevant reference for a client’s problems. We also use generative AI for knowledge search tasks, such as identifying who has previously solved a particular global problem, clarifying specific roles and responsibilities in our contracts, or addressing various IT support queries.

For us, generative AI is about using the knowledge models for faster answers for our people to better serve our clients. This includes using generative AI to identify the riches of data contained in content, documents, and articles, for example, that our people can access to better address our clients’ challenges.

We recently introduced an AI-powered tool, SynitiSense, aimed at enhancing efficiency and productivity. This tool integrates OpenAI’s GPT-4 LLM with vector databases to streamline access to a wide range of internal resources, including support articles, product documentation, sales presentations, and case studies. The goal is to simplify and expedite the process for employees to find necessary information. As the tool is used more frequently, it adapts and improves, enhancing its capability to provide relevant information.

How do you predict data management practices will shape the evolution of enterprise resource planning systems in the coming years?

Our experience with enterprise resource planning implementations and digital transformation projects has revealed several key challenges:

  • Over 40% of go-lives are delayed due to data issues.
  • A significant number of go-lives fail to achieve their intended business objectives because of low-quality data.
  • Current lift-and-shift approaches often lead to dissatisfaction with data quality results.
  • According to a study we conducted, less than 35% of cloud migrations are successful, largely due to data-related issues.
  • No one ever said, “I started cleaning my data too early.”

These challenges indicate that existing approaches are not working, and a new approach is needed.

We foresee a data-first approach taking centre stage in the coming years. Data first realises that businesses need to start data work from day one, long before a project starts. With modern approaches to data management, from day one itself, businesses can start to pit their data against the initial target data model of the target system — all before any global design, process work, or any IT work begins.

By adopting this approach, businesses will be well-placed to identify any gaps in existing data, including missing data, duplicates, incorrect format, and so on, and start making the necessary revisions to ensure that the data is accurate from the start.

This is an approach that puts data first, meaning businesses use data to drive their transformation and inform the rest of its processes. Based on our experience, it is the proven way to get businesses to 99.99% data quality, which subsequently results in on-time go-lives and delivers the intended business outcomes.

What are the most exciting developments from your R&D labs right now?

We’ve shifted our focus from unstructured data to structured data and technical context. To that extent, we’re exploring the use of generative AI to improve data quality through rule suggestions and implementations focused on actual business process outcomes. Our approach differs from traditional methods as we use private model access and training, tailored to our customers’ systems and expertise, to ensure the confidentiality of proprietary information. Ultimately, this will help speed up complex data migration projects — be it in on-premises, hybrid, or multi-cloud scenarios. Effectively, we’re combining traditional lift-and-shift data migration options with advanced data quality transformations to enhance the speed and quality of data.

Additionally, we’re also building our capability to support clients with data projects across different environments, be it on-premises, various private clouds, and public clouds. With our advanced data management module (ADMM), we seek to help big companies integrate these different environments. This integration caters to the unique requirements and maturity levels of different business units within these organisations.

With your experience coaching your kids’ sports teams, what are the parallels between that and leading an enterprise? How do you use lessons learned from one field to drive success in the other?

There are many parallels between coaching a sports team and leading a business, and I apply many principles from my experience with sports in running Syniti.

First and foremost, leading in both fields requires three things: action, teamwork, and leadership. I’ve always been a team advocate, and anyone who knows me probably thinks my middle name is action. That is why my two mottos — be it in sports, life, or work — are:

  1. Think big, be curious, and take action.
  2. We are better and stronger together.

With the above in mind, I will share some of the parallels and learnings in the two fields:

  • Teamwork: As a leader or coach, one of the core elements of building a powerful and successful team is taking the time to identify the right talent. In sports, we try to do different drills and practices to tease out talent or strengths and weaknesses of players.

    In much the same way, various activities, interactions, projects can spark creativity and talent within the enterprise. The pandemic provided me the opportunity to do this more consciously — it’s changed the way I see people. I find myself looking at the whole individual, being more conscious that they bring their whole selves to the field or the business meeting. This allows me to place people in front of the right opportunities.
  • Action: This is more nuanced than simply scoring a goal or making a sales pitch. A crucial element of taking action involves building the ability to respond to changes in the environment. This is where the different drills or interactions come into play; it helps build resilience to change, or to put it another way, “get comfortable with the uncomfortable”.
  • Leadership: Coaching sports gave me a lot of valuable lessons for leading a business. One thing that I took away from my experience in sports is the temperament and conviction of a leader. A coach cannot get caught up in emotions and stresses of the game. We have to maintain a steady head on our shoulders to lead a team and deliver the right advice and strategy to succeed on the field. It’s the same in business: We have to maintain that sense of calm and collectedness.