At its Data + AI Summit, Databricks unveiled Lakehouse//RT, an analytics engine claiming response times as low as 10 milliseconds for smaller workloads. This engine delivers real-time query performance directly to lakehouse environments, unifying operational and analytical workloads on a single data copy within a data lake. Such performance will fundamentally impact how enterprises scale AI in 2026.
Traditional data architectures typically separate operational and analytical workloads, creating latency and complexity. However, the imperative for enterprise AI demands real-time, unified access to all data sources. This separation hinders the speed and accuracy required for AI-driven decision-making and limits advanced automation.
Companies failing to adopt unified Lake Transactional/Analytical Processing (LTAP) architectures and agentic AI tools risk significant competitive disadvantage. Competitors can now make real-time, data-driven decisions and automate complex processes with sub-100ms latency, as demonstrated by Databricks' Lakehouse//RT and Omnigent.
Addressing Data Sprawl and Driving Real-World AI Impact
- Mastercard addresses data sprawl across multiple clouds by using multi-engine catalog federation to access data governed in one catalog, such as Databricks Unity Catalog, from compute engines using a different catalog like AWS Glue, according to Tech Times.
- Workday has re-examined entire business processes and deployed AI agents in production, redesigning operations around agentic systems, reports BizTech Magazine.
Mastercard's and Workday's experiences reveal the persistent challenge of data fragmentation and the transformative potential of agentic AI when data is accessible through unified platforms. Organizations must choose between maintaining complex, siloed data environments or moving towards integrated systems that enable real-time automation.
Unlocking Agentic AI with New Tools and Open Standards
Databricks introduced CustomerLake, an agentic customer data platform (CDP), to transform marketing from static campaigns to agent-driven engagement loops, according to SiliconANGLE. This tool aims to shift businesses from reactive data consumption to proactive, autonomous action in customer interactions.
The company also introduced Omnigent, a meta-harness designed for agentic supervision and control of language models. This innovation allows for human-level oversight of AI agents, facilitating critical business decisions on fresh data with enhanced reliability and accountability.
Databricks and the Linux Foundation launched OpenSharing, an open, vendor-neutral protocol for sharing AI assets like agent skills, ML models, and unstructured data across organizations and platforms without copying files, as reported by Tech Times. This initiative supports a future where AI capabilities are portable and composable across platforms, rather than siloed.
Databricks is constructing an ecosystem for agentic AI, from development to cross-organizational sharing. This strategy aims for more integrated and flexible AI deployment.
Navigating Openness and Unification in Data Architectures
Databricks pushes OpenSharing as an open, vendor-neutral protocol for AI assets. Yet, many other announcements, including LTAP, Lakehouse//RT, Omnigent, and CustomerLake, are specific Databricks products. This creates a strategic tension between promoting an open ecosystem and solidifying its proprietary platform's dominance. The company appears to be creating a 'walled garden' that still relies on its core offerings, even while advocating for open standards.
Databricks' LTAP promises to unify operational and analytical workloads on a single data copy. However, Mastercard's experience highlights the persistent challenge of data sprawl across multiple clouds requiring multi-engine catalog federation. While Databricks offers a vision of unification, the reality for many enterprises is a complex, multi-vendor environment that will require significant effort to consolidate, even with new architectures.
The strategic push by Databricks towards open, vendor-neutral sharing of AI assets via OpenSharing signals a future where competitive advantage shifts from data hoarding to the seamless, secure exchange of intelligent capabilities across organizational boundaries. Enterprises must rethink their data governance and collaboration models. Conversely, businesses that retain separate operational and analytical data choose a significant time lag, preventing them from leveraging agentic AI for proactive customer engagement and real-time operational redesign, as exemplified by Workday.
What Future Operations Will Agentic AI Enable?
The combination of real-time data access from LTAP and Lakehouse//RT with agentic AI tools like Omnigent and CustomerLake enables new, autonomous business functions. This moves beyond process acceleration to fundamental operational redesign. The sub-100ms query performance, coupled with agentic supervision, means AI agents can make critical business decisions on fresh data with human-level oversight. This blurs the lines between automated and human-driven operations. The shift from reactive data consumption to proactive, autonomous action, exemplified by CustomerLake and Workday, will fundamentally change how businesses interact with data and customers, moving towards predictive, self-optimizing systems.
If enterprises effectively navigate the tension between open standards and proprietary platforms, the integration of real-time data architectures with agentic AI tools will likely redefine operational efficiency and competitive advantage in the coming years.
