Across
- 3. The disciplined approach of automating everything that can—and should—be automated.
- 5. A dedicated non-production environment used to stress-test applications before they go live.
- 6. The practice of linking an AI model to a verifiable "source of truth" (like a document library) to prevent errors.
- 8. The ability to see exactly why and when an AI made a specific decision within a workflow for audit purposes.
- 9. The phase where 90% of AI costs now live; the "shock" to the Opex budget.
- 10. Models trained within a company's secure boundary to prevent IP leakage to public LLMs. (2 words)
- 11. A 2026 security posture that uses AI to block threats before they reach the network.
- 13. The old "on-prem" systems that are the hardest to orchestrate.
- 15. Teams where business and IT experts work together on a single project.
- 16. The programmatic "limit" set within a process to ensure an AI agent doesn't exceed its authority.
- 17. The "contract" that allows two software programs to talk to each other.
- 18. The degradation of an AI model's performance over time as real-world data changes.
- 19. The risk of business units deploying unauthorized models, bypassing IT governance. (2 words)
- 22. What a Data Lake becomes when it lacks the governance and connectivity of a Fabric.(2 words)
- 27. The basic unit of text that an LLM processes.
- 28. AI that moves beyond "chat" to independently use tools and execute multi-step business processes.
- 29. The "management" style suited for complex, unpredictable work where AI acts as a co-pilot.
- 30. What AI must be able to redact automatically from documents to ensure privacy (abbr.).
- 31. The consistent data transfer speed you get when you aren't competing for bandwidth with "noisy neighbors."
- 33. The SaaS architecture where each customer has their own dedicated server instance and database, used by Appian. (2 words)
- 37. The "G" in ESG that keeps AI and data within ethical and legal rails.
- 38. An ecosystem of autonomous entities collaborating to solve complex goals.
- 39. The traditional, isolated department that BOAT aims to break down.
- 40. The ability to scale your dedicated resources up or down based on your specific enterprise demand.
- 42. The architectural layer that unifies silos without moving data; the "fuel" for accurate enterprise AI. (2 words)
- 44. When an AI confidently provides a false or nonsensical answer;
- 46. The "deep see" into AI-driven processes to understand why a decision was made.
- 47. The ease of moving workloads because your data and configurations are neatly contained in a single instance.
Down
- 1. The term for generating code via AI without rigorous architectural oversight. (2 words)
- 2. The "O" in BOAT, Gartner’s latest category introduced in late 2025 to unify fragmented automation tools (like RPA, iPaaS, Low-Code, AI Agents) under a single platform for coordinating complex business processes; the coordination of complex systems and workflows.
- 4. The common industry term for AI pilots that never reach production (e.g., "Pilot ____").
- 5. One of Gartner’s 2026 themes focusing on the protection of digital and brand value.
- 7. Logic-based execution required for BOAT workflows where "probabilistic" AI outcomes are too risky.
- 12. The "interest" paid on sub-optimal technology chosen for the sake of speed. (2 words)
- 14. The safety path in a workflow where an AI agent hands off a complex or high-risk case to a human expert.
- 19. The full set of technologies used to build or run an application.
- 20. The primary security benefit of single-tenancy; preventing "noisy neighbors" from impacting your performance.
- 21. The result of "insufficient guardrails," with Gartner predicting 2,000+ AI-related claims by 2026.
- 23. Moving from "Is it working?" to "Is it profitable?" (The 2026 shift).
- 24. The technical "memory" of an AI session; expanding this reduces the need for risky permanent data storage. (2 words)
- 25. The experimental phase that Gartner says 40% of organizations fail to move beyond.
- 26. The high-level security certification often required for government-grade single-tenant clouds.
- 32. The instruction that, if poorly engineered, leads to data leakage or "injection" attacks.
- 33. A type of "Smart Search" that understands user intent and context, not just matching keywords in a database.
- 34. The next wave of AI that doesn't just chat, but takes action autonomously.
- 35. the architecture used to ground LLMs in private company data (abbr.).
- 36. An architecture made of modular, interchangeable "building blocks."
- 41. The ability to move an IT solution from one department to the entire global enterprise.
- 43. A vast pool of raw data that often becomes a "swamp" without governance. (2 words)
- 45. The core entity in a process platform that can now include "Documents," allowing AI to query them like data.