{
  "id": 13096337,
  "title": "Moving AI from pilot to production starts with the data",
  "url": "https://urgent.news/2026/10/09/moving-ai-from-pilot-to-production-starts-with-the-data",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-09T10:30:07.000Z",
  "source": {
    "name": "TechRadar",
    "slug": "techradar",
    "url": "https://www.techradar.com/pro/moving-ai-from-pilot-to-production-starts-with-the-data"
  },
  "original_language": "en",
  "account": "Adoption of AI tools in business functions has surged, with 88% of surveyed organizations using them regularly, according to McKinsey's 2025 global survey. Yet, only 7% claim AI is fully scaled across their organization, and 30% still rely on piloting. Moving AI from pilot to production can be challenging, especially for marketing teams faced with complex data stacks. The issue often lies beneath the model itself, as marketing data is scattered across various platforms, warehouses, and internal systems, each with distinct structures and definitions. During a pilot, the model functions well due to controlled data and narrow scope. However, production introduces new complexities, such as encountering multiple versions of the same metric with different field names and calculation rules. Without a governed understanding of the enterprise environment, the model may make assumptions, leading to inaccurate insights. To bridge this gap, marketing teams need to establish canonical definitions for key concepts and map them across different platforms. They also need to capture the rules used for performance measurement, ensuring AI operates within the same framework as human interpreters. This knowledge layer can be built on top of existing data warehouses, avoiding costly infrastructure replacements. Furthermore, AI requires context, which changes as business circumstances evolve. Factors such as budget changes, promotions, or live market tests can impact performance. Separating persistent marketing knowledge from context specific to each investigation helps AI make accurate conclusions. Collaboration between marketing and data teams is crucial, as marketing leaders understand campaign objectives and performance interpretation, while data leaders comprehend the systems, structures, and controls governing the information. By agreeing on definitions before embedding them in automated workflows, organizations create a stronger foundation for AI. Ultimately, enterprise AI adoption hinges on building confidence in the production environment, rather than continuing to rely on pilots. Ensuring relevant metrics are consistently defined, AI understands inter-system relationships, and it possesses knowledge of current business decisions shaping performance is essential. A well-structured knowledge layer serves as the missing connection between enterprise data and AI reasoning, enabling models to operate reliably and providing businesses with measurable value from their AI investments.",
  "summary": "Turning AI projects from the pilot stage to delivering business value depends on having a strong data foundation.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}