Blogs | April 23, 2025
Artificial intelligence projects are best positioned for success when companies enter into them with an end-to-end activity mindset. Businesses that adopt private AI services have a better chance of covering all the bases, future-proofing their investments so that they are more adaptable to new large language models (LLMs), growing data stores, additional business processes, changing governance requirements and more.
Optimizing your spend profile for implementing your enterprise AI project should be a critical component of any strategy. To avoid unexpected large bills from a cloud provider as their data sets and models grow, businesses will find that using a private AI service that efficiently accommodates scalability, at the right cost and performance levels, is an important design goal.
Successful enterprise AI solutions start with the mission
Business leaders need to solve business problems. Unless AI initiatives are aligned to core business goals from the start, they’ll be hard-pressed to fund them beyond the proof-of-concept stage.
Only 26% of companies have developed the necessary capabilities to move beyond proofs of concept and generate tangible value. At least 30% of generative AI projects will be abandoned after proof of concept by the end of 2025.
Leaders need hard facts: a clear definition of how the effort can support these requirements; plans detailing how AI will be scaled across enterprise business processes; and the expected return on investment.
As companies explore how to maximize AI, they should focus on key milestones: deploying AI models with automation, CI/CD integration, and enabling real-time monitoring; embedding AI solutions across departments; ensuring regulatory compliance, transparency and accountability; and ultimately driving a measurable impact in revenue, cost savings and customer experience.
DXC’s Private AI for the Enterprise can stand behind all these milestones, providing an end-to-end road to adoption, administration and adaptations. Leveraging our framework, you’ll be assured of a business-aligned approach to AI, security for AI data models and applications, strong governance and risk mitigation, and the ability to scale and future-proof investments.
Conducting ideation sessions among all employees to solicit and curate ideas will help developers focus on areas of work that are most important from an enterprise perspective. Once enterprise AI projects are settled upon, activity can shift to determining whether the organization has the right data available and if it is in a usable state — and if not, how to get it to that point. Collecting, harmonizing and pipelining data in a secure way from a company’s various pockets is a lot harder than it sounds. You’ll likely need to conform to various government regulations around data privacy, especially if it moves across borders, for example.
Equally important, you have to assure that your private AI training data and intellectual property — whether it sits behind your firewall in an on-premises data center or co-location facility, or an Amazon S3 bucket — isn't being trained on by public, unsecured commercial models or hosting providers. Additionally, your cloud costs can quickly spin out of control as your enterprise GenAI solutions generate more data.
Intelligent automation is a key AI component of intelligent operations. One DXC customer, a large telecom company, deployed intelligent automation to automatically tackle multiple recurring problems. By automating processes, it boosted patching compliance across its servers and databases from 92% to 95%. That means thousands of IT tickets no longer need to be handled manually by our teams or theirs.
Maintaining your AI architecture
Many businesses struggle with monitoring and managing evolving data and models. This often leads to undetected model drift, gradually degrading the quality of AI responses. Implementing robust evaluation test suites is crucial for continuous quality assurance, as they help test the performance and accuracy of models. Without these test suites, it becomes difficult to assess if a model aligns with business needs.
In the fast-evolving world of intelligent operations, all monitoring and management can take place on a single platform. DXC Technology is helping to bring this about, as we work on delivering centralized management, where companies can get insight into the health of their entire private AI environment, including the overall behavior of their hardware.
AI comes down to the business
All the technical decisions that comprise building AI solutions for enterprise in an end-to-end manner — from how to leverage data, to where to host solutions, to how to maintain applications in an optimized state — will flow from the business goals you settle on.
Your AI endeavor has to tie back to that business framework, and a pilot that moves to production is a success even if it accomplishes just a fraction of the overall goal.
About the authors
About the authors
Holland Barry is the global field CTO for DXC Technology, overseeing technical pre-sales across DXC’s cloud, ITO and cybersecurity business. His team collaborates with partners to deliver innovation and value in customer IT journeys. With 25+ years of experience, Holland has held leadership roles at various tech firms and worked with public and private sector companies on IT infrastructure, automation, networking, and cloud security. Connect with Holland via LinkedIn.
Praveen Cherukuri is a chief technologist at DXC, leading AI-driven digital transformations for global enterprises. With deep expertise in scaling systems, cloud optimization, and AI strategy, he helps organizations accelerate growth and enhance efficiency. Passionate about innovation, he specializes in AI, automation, and data-driven insights—creating competitive advantages for DXC clients across industries.