Contributors
Key Data and AI Trends for 2025
AI and data have been the common thread of almost every trend in 2024. While some companies still struggle, early adopters of AI have progressed faster than expected.
From AI agents and small language models to a human-first approach to AI interfaces and a renewed focus on unstructured data, 2025 promises to be a year of remarkable transformation. As we embrace these innovations, we must focus on driving progress responsibly.
To gain insights into the emerging trends and the transformative impact of AI, we talked with some of the brightest minds from BlueCloud and Coalesce, our trusted partner in the tech space. They shared their perspectives on the big shifts in data and AI in 2025, the challenges businesses must address, and the key skills leadership must develop to oversee data and AI responsibilities.
These are the 2025 AI and data trends leaders need to watch and understand.
Key AI trends in 2025
This year, AI has shifted from pilots to real-world applications. In 2025, it's set to scale across entire enterprises. Generative AI has seen unprecedented adoption, with many companies reaping ROI and expanding use cases. A Bank of America survey highlights 2024 as the year of proving ROI and 2025 as the tipping point for enterprise-wide AI adoption.
AI evolving beyond expectations
The advancements in AI we're witnessing today were almost unimaginable just a year ago. In 2025, AI will be deeply woven into the fabric of our lives, allowing us to experience the world faster, more intuitively, and smarter. The standout use cases will combine AI with domain-specific expertise to tackle complex, high-value challenges.
Bill Tennant, Chief Revenue Officer at BlueCloud, pointed out that transformative innovations often emerge from unexpected areas. "Who could have predicted the impact of Netflix on movie rentals, Tesla on car manufacturing, or privatized companies on space exploration? Similarly, industries like defense and manufacturing might integrate AI more cautiously due to their complexity and sensitivity. Financial services face similar challenges with data sensitivity. Still, the opportunities for AI-driven breakthroughs in these fields are enormous."
Satish Jayanthi, CTO and Co-Founder of Coalesce, emphasized AI's universal impact, saying that no industry will be untouched by its transformative power. "AI will influence everything, from the food industry to drug discovery," he explained. “In consumer-facing applications, we may accept lower precision—like estimating calorie content within a range. But in critical sectors like healthcare, where diagnosing and treating diseases is vital, the margin for error will shrink dramatically."
Tennant also noted that consumer-facing industries are likely to lead the AI transformation. "Many customers are already interacting with LLM-based chatbots and predictive models," he says. "For example, Amazon has integrated LLMs into their search features and other services."
Kerem Koca, CEO and Co-Founder of BlueCloud, believes the biggest transformations will occur in healthcare, manufacturing, and education.
"We will witness groundbreaking advances in drug discovery and diagnostic tools powered by Generative AI. Manufacturing will embrace AI-driven digital twins to optimize operations and reduce waste. And in education, AI will deliver personalized learning at scale, adapting curricula and support to each individual's needs."
AI will be the driving force behind automation's future
AI will continue to shape the future of automation in 2025.
Jayanthi highlighted that while automation can significantly improve data quality, human involvement will remain critical.
"Automation should be the go-to approach for routine tasks, freeing teams to focus on strategic, high-impact work. However, human oversight will remain imperative to ensure the system adapts to complex requirements."
Armon Petrossian, CEO and Co-Founder of Coalesce, added, "AI is simply an advanced form of automation—it's been a core part of our philosophy since Coalesce was founded. AI's role in driving automation will only expand as the market evolves, boosting efficiency and scalability across industries."
Jayanthi also emphasized that automation starts with a mindset. "Every repetitive task should spark the question: Can this be automated?" he noted. With the right tools, this mindset will drive operational excellence.
Generative AI will ignite the next wave of innovation
"GenAI is no longer just a tool for better analytics," Tennant explained. "It's a game-changer for innovation.
We're at a point where AI models are outpacing human capabilities in coding and problem-solving. This allows companies to innovate faster, moving beyond foundational tasks to building scalable solutions.
Tennant added, "Some of OpenAI's models are breaking benchmarks even top-tier coders couldn't achieve. Tasks that once took hours of effort can now be completed in minutes, giving innovators more time to focus on big ideas rather than just laying the groundwork."
Reflecting on its evolution, Tennant shared, "Last year, I thought Generative AI was mainly about improving analytics for structured data. Now, we're merging structured and unstructured data, which opens up new possibilities—and challenges. For example, people can now extract signatures from PDFs and manipulate them, something that would've been unthinkable without advanced training just a year ago."
But with these advancements, Tennant explained, come concerns about costs, performance, and security.
"Businesses don't need perfectly structured enterprise repositories for small-scale use cases to succeed. What they need is accessible, accurate data without unnecessary complexity. In 2025, I hope to see more transparency around successful generative AI use cases."
Koca added that the GenAI evolution will impact companies' technical capabilities and reimagine industries by enabling new business models and redefining value chains.
"One trend that surprised me in 2024 was the rapid commoditization of generative AI tools. While this brought innovation to more people, it also led to a flood of low-value applications," said Koca.
On the flip side, Koca pointed out that he would like to see the "AI hype cycle" fade away in 2025. "AI is not a magic wand—it's a tool that requires thoughtful implementation and measurable outcomes. The focus should shift from "what AI can do" to "what AI is doing effectively."
The RAG approach is set to dominate, driving major transformations
The Retrieval Augmented Generation (RAG) approach—using contextual data to refine model outputs—is quickly gaining momentum. It's making applications like document approval workflows and augmented content generation more accurate and relevant.
Jayanthi explained, "As RAG systems continue to evolve, the ability to integrate multimodal data—like structured data, unstructured text, images, and videos—will be key to their success. Over time, the connection between RAG systems and LLMs will strengthen, making these systems easier to implement. It will open the door for widespread adoption of domain-specific AI applications."
Smaller domain-specific models will unlock greater accuracy
In 2023 and 2024, the spotlight has been on large language models (LLMs) like ChatGPT, Anthropic's Claude, and Meta's Llama. However, LLMs are overkill for many businesses—too expensive, slow, and often too broad for practical use.
Jayanti believes that, in 2025, smaller, domain-specific language models will become the go-to solution for businesses. "AI is becoming ubiquitous, but we're noticing a shift towards smaller, domain-specific language models (LLMs) rather than relying solely on large, general-purpose ones. These smaller LLMs, trained for specific domains like healthcare, allow for more specialized and accurate outputs. This trend will make RAG architecture increasingly important," said Jayanti.
These models are tailored to specific needs, making them more efficient, precise, and easier to train on proprietary data. Smaller models are also ideal for edge and mobile deployments, as seen with Apple's recent mobile AI innovations.
Human oversight is vital to addressing AI hallucinations responsibly
Data quality will be paramount in 2025. With the increasing complexity and variety of data—from structured to unstructured formats—ensuring data accuracy and cleanliness is foundational.
Jayanthi put it simply: "The whole 'garbage in, garbage out' rule has never been more true. If you're building domain-specific models, you've got to focus on accuracy, ethics, compliance, and security from the get-go. And let's be real—human oversight is a must to make sure these systems work responsibly and deliver results."
He also added that it is imperative for organizations to:
- Clean and standardize data: Fixing errors and inconsistencies is key to building a solid foundation.
- Tackle bias and ethics: As LLMs become more commonly used in decision-making, addressing biases and sticking to ethical standards is crucial.
- Focus on strong governance: Good data governance and regular data monitoring are non-negotiable for keeping datasets top-notch.
AI to unleash the next level of human potential
In 2025, AI is set to become an influential collaborator, enhancing human creativity rather than replacing it. The focus is designing AI systems that integrate seamlessly into workflows, offering support when needed without disrupting the creative process.
Jayanthi emphasized how AI-driven advancements in intuitive interfaces and personalized experiences will amplify human ingenuity and empower individuals and organizations to make smarter, more innovative decisions at scale.
"AI should be helpful, not intrusive. At Coalesce, we focus on designing AI that's there when you need it but doesn't get in the way. Nobody wants AI disrupting their workflow or feeling overbearing," explained Jayanthi.
He shared two ways AI can make interfaces more intuitive and natural.
For one, he pointed out that text-based interfaces often work better than voice for tasks requiring precision. "Text allows users to express their needs more clearly, which is crucial for technical tasks. Whether it's asking AI to write code or build an application, text-based input provides the accuracy needed."
Another game-changer is personalization. As AI evolves, it will adapt to individual users' preferences and workflows. "Take an analyst, for example," Jayanthi says. "An AI assistant should tailor its suggestions and insights based on the analyst's unique style and project history. Personalization makes AI more user-friendly and builds trust, turning it into a true partner."
AI will enhance human creativity, not replace it
As technology continues to evolve rapidly, it's unclear how much emphasis organizations will place on human expertise in shaping strong data strategies and how that will impact AI-driven initiatives.
Tennant pointed out that AI's coding skills shine in areas with structured, process-driven tasks. But as these tools improve at automating more work, there will be an even bigger focus on human creativity and the unique qualities we bring to the table. "It will no longer just be about processing data faster but leveraging uniquely human attributes that machines cannot replicate," said Tennant.
Tennant also said that misconceptions about AI's risks and capabilities often arise from a lack of understanding. "As AI continues to grow, I hope more people see it as a tool to enhance, not replace, human efforts. Finding the right balance between caution and curiosity will be crucial in navigating its transformative impact."
Koca believes that AI and automation will redefine how organizations view human capital. The focus will be on hybrid roles—people who understand the interplay of data science, domain expertise, and decision-making.
"Even as algorithms take on more complex data tasks, we'll still need human expertise to define strategies, make sense of AI outputs, and deal with any unexpected issues," said Koca.
To stay ahead of the curve, organizations will need to prioritize continuous learning and empower their teams to keep pace with technology's rapid evolution.
Koca added: "AI enhances creativity—it doesn't replace it. The best uses of AI will always come from humans and machines working together, each bringing out the best in the other."
Key Data Trends in 2025
In 2025, businesses that align their data strategies with clear goals and use AI-driven analytics will be the ones driving innovation, streamlining operations, and achieving sustainable growth.
Data quality, cost, performance, and security will define success
Tennant highlighted three critical concerns for the future of AI: cost, performance, and security.
"AI has endless potential, but businesses need to ensure their data is accessible and secure. Striking that balance will be the key to thriving in the next wave of innovation. Scalable, cost-effective solutions that address performance and security concerns will be the next frontier," explained Tennant.
Another major hurdle for organizations in 2025 will be tackling poor data quality and a lack of automation. Jayanthi highlighted that starting with data governance is key.
"Organizations must identify and prioritize the data that matters most to them," he said.
Jayanti pointed out that assigning accountability is another crucial step. "When businesses empower their teams to take ownership of their data, it means that the data quality is managed by the people who know it best. This approach also ties into the idea of data mesh, which focuses on decentralization and domain ownership."
A robust data foundation will be critical for any AI initiative
Data Mesh can be a game-changer for democratizing data development. It enables domain experts, such as marketing, finance, and sales teams, to work directly with their data and uncover actionable insights while still following governance rules. This approach helps close the gap that otherwise exists when IT solely handles data engineering.
However, Petrossian explained that implementing a Data Mesh framework is not a silver bullet to success with advanced data-driven and AI initiatives, and that a robust data foundation is critically important.
"If data isn't structured properly, it can lead to big mistakes in decision-making. This is especially important with domain-specific LLMs, where even small errors can snowball into bigger problems. To unlock AI's potential and avoid costly mistakes, building and maintaining solid, accurate data foundations will be crucial."
Tennant continued, highlighting a critical issue many companies face: data readiness. "Many businesses hesitate to adopt advanced AI because their data isn't perfectly cleaned or structured. But the truth is, you don't need a huge enterprise system to start small and see success. The key is thinking outside the box."
Companies will proactively embrace data security and compliance
While data democratization is critical, robust data security measures remain a priority in 2025.
"Companies must take a proactive approach to security instead of treating it like an afterthought. Using security frameworks like role-based access control (RBAC) helps by giving people access only to what they need for their roles, so data stays accessible and protected," said Jayanthi.
But how will organizations balance data democratization with robust security and compliance measures? Jayanthi pointed out that Coalesce can help solve the challenge.
"Data democratization is essential, and solutions like Coalesce help lay the foundation for self-service analytics and data mesh implementations," explained Jayanthi. In the same way that Tableau, Qlik, and other business intelligence tools democratized access to data at the dashboard development and consumption level, Coalesce democratizes data development in the transformation layer, enabling teams to transform data efficiently and quickly, all while adhering to strict governance and security rules.
Unstructured data will drive new revenue
Turning unstructured data into revenue opportunities is one of the great challenges and opportunities in 2025.
Tennant added: "It's not just about analytics anymore; analytics are the baseline. The challenge is moving beyond it to innovation—leveraging structured and unstructured data to solve complex problems."
Much of today's data modeling revolves around structured data. However, as we increasingly store unstructured data, the challenge lies in adapting current modeling techniques.
Tennant thinks that the ability to extract insights from unstructured data will be a game-changer.
For instance, retailers can leverage unstructured data to analyze customer feedback and refine product strategies, while logistics companies can use it to improve operational efficiency. The critical factor will be integrating these capabilities into existing business processes to drive measurable impact.
Jayanthi also believes that unstructured data is a largely untapped resource.
"Organizations generate vast amounts of unstructured data, from documents to audio files," he said. "Yet, most haven't fully explored its potential."
For instance, retailers can leverage unstructured data to analyze customer feedback and refine product strategies, while logistics companies can use it to improve operational efficiency. The critical factor will be integrating these capabilities into existing business processes to drive measurable impact.
Jayanthi suggested that the first step is understanding the applications of this data.
“It's not just about data processing but determining its purpose," he explained. "Most likely, organizations will focus on analytics applications." Companies can leverage advanced AI and ML techniques to unlock new revenue opportunities only after defining value.
"In the next 12 to 24 months, I expect to see major shifts in how we approach data modeling and the tools we use to store and process data," concluded Tennant.
Data democratization will scale AI
In 2025, data democratization will also become a crucial factor in scaling AI potential.
Tennant reflected on his journey in analytics and how it mirrors the evolution of business intelligence tools. "In the early days, tools like Tableau and Power BI made data more accessible," he said. "Now, AI is taking this to the next level. We've gone from simple data searches to conversational AI that can reason, interact, and even create avatars."
However, he emphasized that businesses still face challenges in applying these capabilities to achieve a real return on investment. "For individuals, these tools are helpful, but for businesses, they can be exponentially impactful," he noted. "The key is moving from seeing them as a 'nice-to-have' to a 'need-to-have,' which requires a strategic application."
Koca shared his thoughts on balancing democratization and governance. He pointed out that it's all about taking a dual approach: "We need to build trust into the data pipeline with AI-driven governance tools while also creating a culture of accountability." For him, the goal is to give users the freedom to explore data without risking security or compliance.
Key leadership skills for 2025: Empathy, AI mastery, clear communication, cross-disciplinary collaboration
When asked about the critical skills leaders need in AI and data, Petrossian highlighted the importance of foundational leadership principles.
"Designing your data team and influencing your organization's culture is paramount," Petrossian noted.
"It's about more than just technology," he explained. "To achieve success, leaders must strengthen the collaboration between data producers and consumers, built on empathy, open communication, and a clear, shared purpose."
Petrossian also stressed the unique role of leaders in bridging technological gaps through soft skills. "Interpersonal skills are insanely valuable and hard to come by," he said.
Tennant pointed out that tools like ChatGPT democratize coding, but this comes with risks—non-experts might introduce flawed code. He also warned about using open LLM platforms without knowing their security risks, which could lead to data breaches.
"To avoid such scenarios, leaders will need to focus on understanding the business value of AI use cases, ensuring they are executed in the right way. Taking a structured approach to AI integration is essential for managing risks and unlocking its full potential. Companies can leverage advanced AI and ML techniques to unlock new revenue opportunities only after defining value." said Tennant.
Finally, Koca believes that, in 2025, leaders in AI and data must focus on fostering innovation and trust.
--
Recommended resources:
Is Technology Crucial for Successfully Implementing Data Mesh?
How to unleash unstructured data with GenAI
Challenges of AI adoption and how to overcome them, an interview with Bill Tennant
