AI-assisted tools are now integrated across the delivery lifecycle—accelerating code generation, improving test coverage, and enhancing observability and incident response.
As AI transforms how software is built, deployed, and operated, DevOps is evolving through targeted, intelligent automation.
Yet, increasing velocity introduces new reliability challenges. Whether you're provisioning large-scale test data for CI/CD, streaming events across multi-cloud environments, or serving low-latency JSON queries at scale, success now depends on one thing: end-to-end data reliability.
Experts recently joined DBTA for roundtable webinar, DevOps in the AI Era: Fueling Innovation at Scale, to explore how leading teams are combining AI-driven automation with data-centric DevOps practices to deliver software that's not only faster—but verifiably trustworthy.
According to Jatinder Luthra, sales engineering advisor at Perforce Delphix, the machine learning/AI landscape consists of two approaches to AI: Predictive ML and Generative AI.
They both share the same challenge: protecting sensitive data without sacrificing ai/ml innovation.
Masked data preserves analytical patterns, distributions, and relationships while eliminating PII exposure, enabling compliant data exploration across all AI/ML and analytics workflows, Luthra explained.
Delphix is perfect for this occasion as it creates masked clones of production data within Snowflake/Databricks, enabling ML teams to work with production-scale data without PII exposure—while BI teams continue using governed real data for business decisions.
Database teams are under pressure to push boundaries with AI and cloud-native services, while simultaneously navigating a growing web of regulations and public concerns about data privacy, ethics, and security, said Kellyn Gorman, multiplatform database and AI engineer and advocate at Redgate.
For many, the question isn’t whether to innovate, but how to do so without crossing legal, ethical, or operational lines. AI must scale and DevOps must evolve. Governance must be embedded, she said. Real innovation happens when all three are connected. Redgate offers AI that works responsibly, securely, and at scale.
Software development is a system, said Peter Caron CEO at 3T Software Labs. The system is the workflow required to build, ship, and maintain a software product. It is not simply the tasks that change with AI, it is the roles of everyone involved in the decision-making processes along the entire SDLC.
Strategies for AI software development include:
Optimize existing tasks within the existing structures and processes: Ex. using AI to write queries to retrieve data from databases.
Anticipate new tasks on top of existing workflows: Ex. coders using AI tools in pair programming.
Alter the logic of value creation and create new workflows: Ex. different people overseeing workflows and decision-making.
Shift workflows and reconceptualize the organizational system in which the company is operating: Ex. As a system-change e.g. how and at what point decisions are made.
According to Caron, companies may succeed by doing the following:
- Integrate AI agents directly into existing workflows directed by an engineer
- Allow AI to improve coordination between tasks to create new workflows
- Assign Product Operations the task of finding and exploiting leverage points in the new workflows
- Bundle old engineering tasks into new roles
- Expect engineers to spend more time imagining outcomes
Traditional players will use AI to optimize existing offerings and improve current workflows, while advanced companies will make AI the core of their solution and integrate new workflows and redefine engineering roles, Caron concluded.
For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.