The foundational elements of AI architecture that IT leaders need to scale
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Elastic CIO Adnan Adil identifies four enduring AI architecture foundations—data quality, context engineering, governance, and human expertise—that organizations should invest in as models evolve toward agentic systems
- Gartner predicts companies will abandon 60% of AI projects through 2026 if they are not backed by AI-ready data, with industry surveys consistently ranking data quality among the greatest barriers to AI success
- Context engineering, distinct from prompt engineering, designs the entire information environment around a model using RAG and vector databases; Adil says "minimum context, correct and current data, and machine-readable information are critical"
- AI governance gaps cause systems to process more data than necessary, inflating token and API costs, while expanding attack surfaces with risks including prompt-based data leakage, model vulnerabilities, and adversarial inputs
- Elastic's 2026 report finds 85% of IT decision makers expect to enable LLM observability for their internal generative AI apps, with Adil calling observability "huge" for cost control and engineering efficiency
- Deloitte's 2025 Tech Executive Survey shows nearly 70% of respondents plan to grow teams in response to generative AI—a figure Adil positions as a contrast to "widely reported AI-related cuts"
Why it matters: With Gartner forecasting 60% of AI projects abandoned by 2026 without AI-ready data and 85% of IT leaders planning LLM observability rollouts, the piece gives IT leaders a framework for which bets to make when underlying models change every few months. Organizations that treat data pipelines, context design, governance, and human oversight as bolt-ons rather than foundations face wasted spend on agentic AI initiatives that never reach production reliability.
Ask SkimNews
