BUSINESS WHITE PAPERThe Pure Storage Platform for AIPure Storage® accelerates and simplifies AI deployments, enhancing their value to the enterprise.
Uncomplicate Data Storage, Forever2BUSINESS WHITE PAPERWith readily available generative artificial intelligence (GenAI), AI has become a sine qua non of information technology (IT) operations. Enterprises in finance, medicine, manufacturing, transportation, security, and others all realize that AI is now a survival issue for them. Those that use AI to identify trends, make accurate predictions, serve clients faster with less effort, and so forth have distinct competitive advantages over those that don’t. The increasing importance of AI-based solutions makes reliable, easy-to-use IT services a must for production deployments. This brief surveys AI needs throughout the project lifecycle (primarily from a storage perspective) and shows how the Pure Storage® portfolio of storage system, data services, and workflow management products for Kubernetes promote efficiency both for AI and IT infrastructure teams as well as developers and MLOps engineers who design, implement, and run AI applications.The AI Project LifecycleOrganizations undertake AI projects to support mission objectives such as,•More accurate medical diagnoses•Acceleration of genomic research •More predictable market fluctuations•Bank card fraud detection•Rapid identification of security threats•etc.Whatever their objectives, in-house AI projects tend to follow a trajectory similar to that shown in Figure 1, from conception, development, production, to evolution. AI projects generally start with a proposed model (algorithm) and train (refine) it iteratively using steadily increasing amounts of available or easily acquired input data for which outcomes are known until it reliably produces inferences (outcomes) that support the mission objective. For example, a medical diagnosis model might be trained using thousands of MRI scans with known diagnoses. The finished model would then take live scans as input and suggest diagnoses to medical practitioners.Models are typically envisioned by small groups of data scientists, often in functional or business organizations rather than in IT teams. Data scientists start with modest IT resources (e.g., public cloud virtual machines and storage) to experiment with model variations. “The playing field is poised to become a lot more competitive, and businesses that don’t deploy AI and data to help them innovate in everything they do will be at a disadvantage.”PAUL DAUGHERTY, CHIEF TECHNOLOGY AND INNOVATION OFFICER, ACCENTUREQUOTED IN: HTTPS://WWW.SALESFORCE.COM/BLOG/AI-QUOTES/FIGURE 1 AI Project Life Cycle