COMPUTING WHERE THE DATA RESIDES
What’s hot: “Instead of pulling data into different environments, we’re bringing compute to the data,” said Scott Gnau, head of data platforms at InterSystems. “For a long time, the common approach was to move data to wherever the applications or models were running. AI depends on fast, reliable access to governed data. When teams make this change, they see faster results, better control, and fewer surprises in performance and cost.”
Current status: The movement of compute power to where the data resides “is already emerging in industries that are pushing the limits of data volume and privacy, especially in healthcare,” said Gnau. “Many are experimenting with generative and agentic AI, and they’re realizing that performance depends as much on where the data is processed as on the model itself. We’re still early, but the mindset is shifting.”
Potential roadblocks: Legacy infrastructure and old habits stand in the way of close-in compute, said Gnau. “Some organizations still move data around for every task or analysis, which drives up cost and takes more time. Shifting to data-centric compute means thinking intentionally about where work gets done and building systems that can handle data safely at the source.”
ACCESS-CONTROL MODELS
What’s hot: Over the coming year, the rise of agentic AI will force organizations to re-evaluate their access-control models, said James Urquhart, field CTO and technology evangelist at Kamiwaza AI.
Current status: Over the coming year, “A wave of innovations will focus on dynamic security, including relationship and attribute-based access control,” said Urquhart. Within the next 4–5 years, “Organizations that rely heavily on AI will need entirely new security frameworks to monitor agent behavior in real time and contain risks from unplanned actions.”
Potential roadblocks: Organizations need to recognize that “role-based access control and other human-defined access control models are ill-equipped for governance in an AI-driven world, especially with the rise of multi-agent systems,” said Urquhart. “As autonomous agents interact and organize in unpredictable ways, security must shift from static permissions to behavior-based safeguards capable of detecting emergent patterns.”
AI AGENTS SPEEDING UP DATA ACCESS
What’s hot: In 2026, we’ll see AI agents increasingly accelerating the work of data teams as “they build data pipelines, improve data governance, enhance the quality of the data they transform, and overall improve the data maturity of their organizations,” said Michael Andrew, chief data officer at Salesforce. They will replace dashboard or custom applications. “For example, our sales agent is able to put together account briefings and customer briefings in seconds that in the past would have taken them days or weeks.”
Current status: “The transformation is widespread, though the adoption is lagging behind the technology’s potential,” said Andrew. “For many companies, the first application of agentic AI will be in structured domains like software engineering and customer service.”
Potential roadblocks: “Companies need a trusted platform where they bring all of this together with the right security, governance, and ease of development,” said Andrew. “It’s not the time to ‘DIY AI’ as there is significant complexity under the hood with model management, security, service reliability, data management, and many other aspects to create enterprise AI agents that work 24/7 in production.”