Jay Allardyce is General Manager of Data and Analytics at Insightsoftware. He is a technical director with over 23 years of experience in enterprise B2B companies such as Google, Ittake, GE and HP. He is also the co-founder of Genai.Works, leading the largest AI community on LinkedIn.
InsightSoftware is a global provider of financial and operational software solutions. The company provides tools to support financial planning and analysis (FP&A), accounting and operations. Its products are designed to improve data accessibility and help organizations make informed decisions in a timely manner.
You have stressed the urgency of enterprises adopting AI to cope with the rise in customer expectations. What key steps should enterprises take to avoid the trap of “AI FOMO” and adopt universal AI solutions?
Customers are making businesses aware that they want to improve their AI capabilities in the tools they use. In response, companies are eager to meet these requirements and keep pace with their competitors, creating a busy cycle for all involved. Yes, the end result is AI FOMO, which can drive the business to rush to innovate to simply say, “We have AI!”
My biggest advice for a company to avoid getting stuck is to take some time to understand what pain points customers require AI to solve. Is there a process problem that is too intensive? Are there repetitive tasks that need to be automated? Is it possible to easily calculate the calculations through a machine?
Once businesses have this necessary environment, they can start adopting purposeful solutions. They will be able to provide AI tools for problem-solving customers, rather than those that simply exacerbate the confusion that exists.
Many companies are eager to implement AI without fully understanding their use cases. How do organizations identify the right AI-driven solution that suits their specific needs rather than relying on common implementations?
On the customer side, it is important to maintain ongoing communication to better understand which use cases are the most pressing. Client Advocacy Committees can provide useful solutions. But, besides the customer, it is also important to look inside the team and understand how adding new AI tools will affect internal functionality. For every new tool introduced to customers, the internal data team is facing new variables and new data being created.
While we all want to add new features and show them to our customers, AI deployments won’t be successful without the support of internal data teams and post-development scientists. Align internally to understand bandwidth, and then look outward to determine which customer requests can be determined with appropriate support behind it.
You have helped 1,000 Fortune companies adopt a data-first approach. What is the significance of “data-based” for companies, and what are the common pitfalls encountered by companies during this transformation process?
In order for companies to become “data-driven”, businesses need to learn how to use data correctly. A truly data-driven team can perform correctly on data-driven decisions, which involves using information to inform and support business choices. Decision makers rely not only on intuition or personal experience, but collect and analyze relevant data to guide their strategies. Making decisions based on data can help businesses come up with smarter objective insights, which can mean the difference between strategic and impulsive decisions in a rapidly changing market.
A common pitfall to achieve this is invalid data management, which leads to “data overload”, where the team is burdened with a lot of data and cannot do anything about it. When an enterprise tries to focus its energy on the most important data, if it is not managed correctly, excessive IT access can lead to delays and inefficiency.
Given your background in collaboration with IoT and industrial technology, how do you view the intersection of AI and IoT development in industries such as energy, transportation and heavy structures?
When the IoT enters the field, it is believed that this will allow for greater connectivity to enhance decisions. This connectivity, in turn, unleashes a whole new world of economic value, and in fact, it’s the reason for the industrial sector.
The problem is that many people focus on “smart pipelines”, using the Internet of Things to connect, extract and communicate with distributed devices with fewer results. You need to determine the exact problem you want to solve, because you are already connected to 400 heavy construction assets or 40 owned power units. The end result or problem to be solved ultimately boils down to understanding which KPIs can improve which, thus driving top-level lines, workflow productivity, or bottom-line savings (if not combinations). Each business is subject to a set of advanced KPIs that measure operational and shareholder performance. Once these issues are identified, the solved problems (and therefore, which data will be useful) will be clear.
With this foundation, AI (both predictive and biological) may have a 10-50-fold impact on helping businesses improve their work. Optimized supply, truck rolling and service cycles are all based on clear demand signal patterns that match the required input variables. To illustrate, the concept of “at the right time, in the right place” could mean millions of dollars for construction companies – because their inventory levels of AI-based models and optimized service technologies are small, which knows or predicts when a machine may fail or when a service event occurs. In turn, the model, combined with structured operational data and IoT data (for distributed assets), can help companies be more dynamic and slightly optimized without sacrificing customer satisfaction.
You have talked about the importance of using data effectively. What are the most common ways companies abuse data? How to turn it into a real competitive advantage?
The term “artificial intelligence” can be misleading when it comes to face value. Entering any and all data into the AI engine does not mean that it will produce useful, relevant or accurate results. When teams try to keep up with the speed of AI innovation in today’s world, sometimes we forget the importance of complete data preparation and control, which is to ensure that the data that provides AI is completely accurate. Just as the body relies on high-quality fuel itself, AI depends on clean, consistent data to ensure the accuracy of its predictions. This is crucial especially in the world of financial teams, so the team can produce accurate reports.
What best practices are the ability to empower non-technical teams within an organization to use data and AI effectively without overwhelming them with complex tools or processes?
My advice is to get leaders to focus on empowering non-technical teams to conduct their own analysis. To be truly agile, the technical team needs to focus on making employees across the organization more intuitive, rather than focusing on the growing backlog of finance and operations. Deleting a manual process is actually the first step in this process, as it allows operations leaders to spend less time collecting data and more time doing analysis.
InsightSoftware focuses on bringing AI into financial operations. How does AI change the way CFOs and financial teams operate, and what are the biggest benefits AI can bring to financial decisions?
AI has had a profound impact on financial decision-making and financial teams. In fact, 87% of teams have used it at medium to high interest rates, which is an excellent measure of its success and impact. Specifically, AI can help finance teams generate important predictions faster and therefore more frequently – significantly improving current prediction performances, with estimated 58% of the budget cycle time exceeding five days.
By adding AI to this decision-making process, teams can use it to automate tedious tasks such as report generation, data verification, and source system updates, freeing up valuable time for strategic analysis. This is especially important when financial teams need agility and flexibility to increase resilience. Take the financial team in the budget and planning cycles as an example. AI-driven solutions can provide more accurate predictions and help financial professionals make better decisions through more in-depth planning and analysis.
How do you view the need for data development over the next five years, especially related to AI integration and the shift to cloud resources?
I think the next five years will indicate the need to enhance data agility. As the market changes at a rate, data must be agile enough to keep the business competitive. We see this in the transition from on-premature to cloud, where businesses have data, but none of this is useful or agile enough to help them transform. Enhanced flexibility means enhanced data decision making, collaboration, risk management and many other features. But, at the end of the day, it equips the team with the tools they need to effectively meet challenges and adapts to trends or market demands as needed.
How do you ensure that AI technology is used responsibly? What ethical considerations should enterprises prioritize when deploying AI solutions?
In parallel between the rise and adoption of the cloud, organizations are afraid to provide their data to some unknown entity to operate, maintain, manage and protect. It took many years to build that trust. Now, with AI adoption, a similar pattern has emerged.
Organizations must once again trust a system to protect their information, in which case it produces viable information that is factual, cited, and trustworthy. With Cloud, it’s about “who owns or manages” your data. With AI, it revolves around the trust and use of that data and the derivation of the information created by the results. That being said, I recommend that organizations focus on these three things when deploying AI technology:
- tilt – Don’t be afraid to use this technology, but adopt and learn.
- Grounding – The corporate data you own and manage is fundamental truth in terms of information accuracy, as long as the information is true, factual and cited. Make sure you understand how the AI model is trained and what information is used in building your data. Like all applications or data, context matters. Non-driven applications can produce erroneous or inaccurate results. Just because AI produces inaccurate results does not mean that we should blame the model, but rather understand the reasons for the feeding model.
- value – Understand the use cases where AI can significantly improve the impact.
Thanks for your excellent interview, readers who wish to learn more should visit Insightsoftware.
