The following Panel of Experts was organized by the Austin Business Journal in an effort to better understand how AI has changed — and will continue to impact — the business community in Austin. Participating in the panel were David Abendschein, CTO, Consero Global; Cristina Carl, founder and CEO, Ediphi; Matt Ache, president, TyRex Group; Seamus Jones, director, server technical marketing, Dell Technologies; and Matt Wursta, founder and CEO, Wursta.
The discussion was moderated by Jay Boisseau, Ph.D., CEO and co-founder, Vizias; director and founder, Austin AI Alliance; executive director and founder, Austin Forum on Technology & Society.
Jay Boisseau: It’s hard to believe it hasn’t even been two years, and suddenly generative artificial intelligence — gen AI — is everywhere. What can businesses do now to catch up? As we know, it’s not just ChatGPT anymore; we also have multiple large language models (LLMs) now. We have numerous image and audio-video programs. Can small companies use this effectively to grow their business? What should they do?
Seamus Jones: The rate of change has been unprecedented, even compared to when ChatGPT first came out two years ago. Just staying on top of that rate of change is really difficult for customers of all sizes. The biggest thing that needs to happen in the near term is that businesses need to determine what portions of LLMs or gen AI are going to be productive for their specific business. Then it will be a matter of integrating those into your infrastructure. I think AI could be a great equalizer, even for the small business framework, because if they’re able to take advantage of some of the promise AI holds, they can keep themselves competitive, if not leapfrog others within the marketplace.
David Abendschein: I would agree. It comes down to: what is the value you’re trying to create for your customers? It’s very challenging for SMBs to invest in something like this on their own, so Consero is breaking down all of our use cases and doing a deep dive of due diligence to determine what value looks like on behalf of our clients. No matter which direction you go, if you plan to use machine learning for automation and data analysis, or generative AI for improved interaction with your data, it will cost money to implement and maintain. There’s a cost associated with data consumption, so you have to understand what your ROI is going to be. I’ve heard some horror stories of companies just implementing AI with no real strategy. Their data consumption costs skyrocketed and the expenses really outweighed the benefits that they were expecting to see.
Cristina Carl: It’s incredible how quickly gen AI has become part of our daily lives. For small businesses wanting to catch up — or even get ahead — I’d suggest starting by immersing yourself and your team in what’s out there. Training can be a game-changer because it builds internal expertise and gets everyone thinking creatively about applications. However, consultants can not only speed up adoption but also offer new ways to think about things. We need to consider not just enhancing our current workflows but rethinking how work gets done from the very first step.
Boisseau: You still need to know your business problems, though. You need to have enough AI literacy at your executive level, so you know where AI could be useful in solving them. Consultants could help with this for sure, but your long-term goal is to have that expertise in-house, right?
Abendschein: I can speak from experience on this one because we did bring in some outside consultants to help. One group came in to train our leadership team around how AI might be useful in certain situations. Another group came in to help with the actual implementation. This approach allowed for a templated model for us to both accelerate and eventually bring in-house.
Matt Ache: Jay, going back to your question about small businesses in particular, I can say that in manufacturing, AI tools will favor smaller businesses with flexibility, entrepreneurial spirit and an understanding of the need for speed in all phases of manufacturing. The most effective way to utilize and experience AI in manufacturing is to have an open mind, and not have a departmentalized business structure. This means you need to include every employee and department, and make sure they are integrated into your manufacturing business design and development.
Matt Wursta: Gen AI has exploded onto the scene, and it’s changing everything. But as Seamus suggested, it’s a real opportunity for small businesses — not just to compete, but to lead. At Wursta, we’re helping businesses to do just that. I would say there are four ways small businesses can get ahead. First, focus on your needs. Don’t get distracted by the shiny new toys. Next, get your data ready. AI needs good data to work its magic. Make sure yours is organized and accessible. Third, empower your people. AI like Gemini for Google Workspace is a powerful tool, but it needs skilled people to wield it. Finally, don’t be afraid to ask for help. We understand that navigating the world of AI can be daunting, especially for small businesses.
Boisseau: So drilling down on that a bit further, what are some of the best AI use cases to emerge thus far from the small business world?
Jones: I think one of the key functions for AI at the moment is to eliminate some of those tedious, menial tasks no one wants to do — or ones that may have a high cost associated with them. If you talk to any IT administrator, they’ll tell you what a pain it is to evaluate IT logs. It can be absolutely mind-numbing. So, one use case Dell developed for a customer involved sorting through logs, and then being able to interact with it as an IT administrator. So through agentic AI, we can set up a framework that will actually resolve some of those issues.
Boisseau: So I assume, in that case, you’re using the gen AI’s capabilities to interpret the question, but not to calculate the answer, correct?
Jones: Yes. And in some cases, the AI can also suggest to the user “these are the possible fixes based on the fastest time to resolution in other cases.” So, we’re essentially using the database of other cases to find a resolution.
Wursta: Once again, I agree with Seamus — automating processes can free up so much time, and this is especially impactful for small businesses, where everyone tends to wear so many hats. Think of the time savings we could enjoy in regard to tasks such as scheduling or managing invoices, for instance. AI can analyze customer data to create targeted marketing campaigns. This can increase the effectiveness of marketing efforts, which can provide critical support to small businesses as they compete with larger businesses with bigger budgets.
Carl: You know, AI is already proving to be incredibly useful across a variety of functions. In customer service, chatbots can handle routine inquiries around the clock. In the back office, AI-powered tools for accounting and financial management are streamlining operations and cutting down on errors, which saves not only time, but money. Additionally, AI is making onboarding and training more efficient. By customizing training materials to each individual’s role and learning style, AI ensures new hires can hit the ground running. Ultimately, AI will not only enhance current workflows but also rethink how work gets done from the very first step.
Ache: I’ll use our own example, if I may. TyRex is focused on combining additive manufacturing and 3D printing with AI. AI can be a game changer for manufacturing, honestly.
Boisseau: Some of you mentioned some highly personalized interactions with AI, which brings me to my next question. How can businesses be sure their data won’t be misused when they import it into an AI model? So whether they’re training an AI model — or using something like retrieval-augmented generation (RAG), for instance — how do you make sure private data is protected? What are the risks, and how do we mitigate them?
Abendschein: To do anything AI-related, companies need a data strategy and strict data management practices in place first. Consero handles confidential and sensitive financial data across our clients; people need to supervise the quality of the data inputs and outputs because it’s very sensitive information that needs to always be 100% correct. Internally, we also have strict guidelines around the usage of ChatGPT, and the data has to be anonymized for anything done within the public model.
Boisseau: Exactly. If you’re using a public model, don’t put sensitive data into it.
Wursta: We partner with Google, because they’ve built a reputation on security and trust. With Google AI tools, including Gemini, your data is kept separate, and you retain full control. It’s not used to train gen AI models without you choosing explicitly to opt in and grant your permission.
Jones: That’s good to know. At Dell, we always say data has gravity. Not all data is created equally, right? We’ve seen a lot of customers implementing on-premises deployments for sensitive data, and then also using cloud infrastructure and things like that for other use cases. So they’re using this hybrid approach to their AI deployments right? A balance like that is ideal.
Carl: Data privacy really is a huge concern when implementing AI, with risks ranging from breaches of sensitive data to the misuse of proprietary information. Having the right guidance makes all the difference in navigating these challenges. It is critical to select platforms and vendors that prioritize security. As Seamus said, a balanced approach is key.
Ache: For our part, TyRex has searched the internet for nearly 200,000 product ideas and placed them into a safe, secure and sanitized AI data box that we call the Cosmic Inquiry. Essentially, it has security boundaries, and access to the data box is under our control.
Boisseau: How do you see AI changing the economy in Austin, and more broadly, Texas? Most of the core work seems to happen in the Bay Area. Is there room for Austin to be a leader in this technology? What would that take?
Carl: Austin is already showing signs of becoming a hub for AI innovation. With its strong tech community, entrepreneurial culture and world-class universities, Austin has the potential to take a leading role. It’s about leveraging our local strengths — such as the collaborative spirit between academia, startups and established companies — and focusing on creating an ecosystem where innovation can thrive.
Jones: I agree. The tech sector in Austin is $469 billion of our economy. That’s over 20% of the GDP of Austin, so it’s huge. I think if we look at the historical data of the area, we have this marriage between different creative mindsets. We started as a big music city, and that led to the development of SXSW, which now has a scope well beyond music, right? In general, we have a lot of innovation happening, both in startups and large tech firms. So we have this confluence of talent associated with these different things, and that’s why people are investing in the Austin area. That’s why we’re being called Silicon Hills. This year Dell celebrated its 40th anniversary. Part of the company’s success over the years can be attributed to the strong talent pool of innovators and technologists here in Central Texas.
Abendschein: Cristina mentioned academia. If we continue adding to the existing AI-focused courses and instructional capabilities within UT, it will push Austin into the forefront of the AI revolution. The cost of living is a little bit better here than in the Bay Area so that, along with Texas being a pro-business state, will continue to drive tech firms and talent into the Austin market.
Wursta: Nearly half of all Austin residents hold at least a bachelor’s degree. The appeal of having that highly educated talent pool really can’t be overestimated.
Ache: We actually believe Austin — and Texas as a whole — can be leaders in this new era of manufacturing as well, by bringing these types of businesses and jobs back to the area. In fact, I will take it further and say that with the help of AI, I believe we can become leaders in manufacturing, even worldwide.
Boisseau: As we know, AI is increasingly demanding more data centers, which consume huge amounts of energy. Texas officials have said the state will need to double its electric output within six years to keep up with the overall demand. What’s being done to reduce power usage? Where are the business opportunities in finding lower- and cleaner-energy solutions?
Carl: Energy consumption will be a significant challenge as AI grows, but it’s also an opportunity. Many data centers are working on becoming more energy-efficient through advanced cooling systems and better hardware design. Innovations in areas like energy storage, smarter grid management, and sustainable infrastructure can not only help meet the demand, but also create new industries and jobs.
Abendschein: I was recently out at the United Nations General Assembly in New York, and at the Wall Street Journal house. They had a three-day session with speakers that they brought in, and this was a major topic. They interviewed an energy firm there, Constellation Energy. They are bringing back nuclear facilities and even went into some detail around building out micro-nuclear facilities that could be built alongside these data centers that consume large amounts of energy, and it’s clean energy. I believe Microsoft just signed a large deal with Constellation Energy, who just brought a unit on Three Mile Island online to utilize nuclear power to generate one of their LLM data centers.
Jones: Jay is correct. ERCOT estimates that by 2030, Texas is going to need 152 gigawatts of energy during peak performance hours. So, when we talk about sustainable energy in Texas, that is going to require all forms of energy to be part of the grid, right? Because right now, there are companies choosing not to invest in Texas, because they’re worried about scale. At Dell, we’re focusing on enabling customers to be able to make the best use of their energy by reducing emissions from purchased goods and services by 45% by 2030, and from sold products by 30% by 2030.
In terms of the product development life cycle, we know we will need to consistently deal with increased power demands of these systems, so that means we need to focus on the best performance tokens per watt. So we’re challenging our engineers to develop more platforms that have thermal profiles to ensure more and better control of the systems. That way, if you’re not necessarily pushing a system to max utilization, you can set thresholds.
Boisseau: Texas lawmakers are discussing possible AI regulations, as are most other states. What’s your view on how local, state and federal governments could reduce potential risk to IP and the spread of misinformation — or perhaps more importantly, disinformation? Can we ensure safety, without hindering innovation?
Jones: In terms of government regulation, I think they should really be focusing on building up a robust infrastructure. Regulation shouldn’t be a hindrance. It should be a stem for innovation.
Ache: Agreed. Providing boundaries and ethical guidelines are a must, but laws and legislation should not put constraints on the creativity and innovation of AI minds.
Carl: It’s a delicate balance between regulation and innovation, for sure. I think it’s important for governments to work closely with industry experts, technologists, and ethicists to create thoughtful regulations that protect people, without stifling creativity. Self-regulation within industries will be critical, too. Ultimately, clear and flexible regulations — crafted with input from a variety of stakeholders — will help keep innovation alive, while ensuring AI is developed responsibly.
Abendschein: There always needs to be a level of regulation, especially with something as powerful as AI. One of the larger challenges around gen AI is whether or not there is any bias in the output generated from the LLMs. People want to feel confident that the responses are being analyzed objectively, and not moving in one direction or another.
Boisseau: I’ll add a little commentary here, because we’re setting up a tech summit on disinformation and trust here in the Austin AI Alliance with UT and some other folks. I think disinformation is the greatest problem of our time. I’m less concerned with bias, because there are valid personal and political views on some issues that are all fact-based on all sides. But I’m absolutely against disinformation — in other words, spreading wrong information with malicious intent. This is where AI and fact-checking can potentially be really useful.
Jones: As a parent, I realize we’re training a new generation of people to use these new tools responsibly as well. So that is something to think about.
Boisseau: Absolutely. So what do you believe are the most promising areas for AI?
Carl: While AI’s potential is vast, a few areas stand out. In health care, for instance, AI assists with diagnostics and predicts patient outcomes, saving lives and enhancing care quality. Additionally, I’m excited about AI’s role in supporting underrepresented groups — like immigrants and new American workers — by offering multilingual learning tools and personalized training. This not only meets their specific needs but also helps build a more diverse and skilled workforce. However, I believe we’re just scratching the surface; there are countless use cases we haven’t even begun to imagine that could revolutionize industries and everyday life.
Ache: Cristina, I agree breakthroughs in health care could provide big areas for growth — especially medicines and medical devices, sourced from a better understanding of molecular structures and individual cell technology. The medical research community will thrive on AI’s huge data processing capacities and capabilities.
Wursta: Absolutely. In my opinion, AI actually has the most potential to help in ‘non-hard-tech’ science — medicine, but also math, physics and materials science. AI is already helping to completely change the velocity of research in these areas.
Abendschein: In finance and accounting specifically, there is a tremendous opportunity for artificial intelligence through the analysis of data, providing forward-thinking business insights, forecasting, predictive analytics and comparative analytics. There will also be a push to automate much of the accounting and finance vertical so we will see everything done at a much faster pace, with improved data quality and enhanced security controls around sensitive data.
Boisseau: With AI moving as fast as it is, how do you make sure that you’re staying on the cutting edge? What tools and resources do you use to keep up?
David Abendschein: I actually wasn’t aware of the Austin AI Alliance, but I’d love to join because it sounds like a great opportunity for like-minded folks to get together and talk about their experiences. I’m focused primarily on the engineering side, but attending these different alliances and meetups allows you to talk about all the different parts of the ecosystem. Sometimes we get too siloed in our environment, but it’s important to have a handle on the big picture.
Seamus Jones: You really can’t beat face-to-face interaction. We’re lucky at Dell. We’ve got a bank of experts that work with us, so we have an internal resource around that framework of talent and expertise. The big thing, though, is understanding the TLDRs of the world. Those forums where we can get summaries of exactly what’s happening in the marketspace are so beneficial. It’s like when you’re driving down the highway. If you’re traveling at a faster speed, you need a further point of reference. Right now we’re going at a faster speed than that knowledge gain, so we’re looking further out on the horizon.
Boisseau: Manufacturing is certainly an industry where we’d expect to continue to see technological advancements in the world of AI — Matt, can you expand on this for us?
Ache: Our TyRex’s Voyager software was created out of necessity, because a manufacturing business’s product development pace must equal Technology Time (SM) in order to stay competitive. This software helps create a constant stream of new product ideas that can be sourced, researched, developed and prototyped in a very time-efficient and cost-effective manner. At this time, we provide the new product initial search functions of this software for free to anyone who takes our two training courses. With this software and structure, we can reduce our customers’ additive manufacturing product development time and cost by 60-80% or more. So, it’s pretty exciting.
Boisseau: So to wrap things up — if you had to guess, which advanced applications of AI do you think will be mainstream in five years?
Jones: It’s difficult to predict, but I think quantum computing could come to light. There’s been a lot of promise out there for a while, but in five — maybe 10 — years, quantum will become more pervasive in the marketspace.
Abendschein: Because of power consumption issues we discussed earlier, and with more and more people being concerned about our carbon footprint, I think there will be drastic improvements in the climate space, due to the impact of AI across the globe.
Carl: I agree, David. I think AI will be critical in addressing some of society’s most complex problems. On the environmental management front, AI could help us predict and mitigate risks — whether that’s through advanced climate modeling or managing natural disasters in real-time.
Boisseau: Let’s close with this. I hope that in five years, what we’ll be seeing is an improvement — not just in the applications — but in the methodologies which feed the applications. I think we may begin to see some truly cognitive methods make their way into AI. The technology is very crude right now, but I think agentic AI will eventually enable you to have a semi-digital, cognitive twin — a really smart one, that makes much better decisions for you. You might even release some of them to be autonomous — if you dare.
This article originally appeared on Austin Business Journal. To learn more about how Consero is harnessing the power of AI to provide unrivaled value in financial services, connect with us here.
