{
  "id": 14664271,
  "title": "What it takes to move enterprise AI beyond the chatbot",
  "url": "https://urgent.news/2026/10/15/what-it-takes-to-move-enterprise-ai-beyond-the-chatbot",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-15T09:13:59.000Z",
  "source": {
    "name": "TechRadar",
    "slug": "techradar",
    "url": "https://www.techradar.com/pro/what-it-takes-to-move-enterprise-ai-beyond-the-chatbot"
  },
  "original_language": "en",
  "account": "Enterprise executives have long been told that AI will revolutionize work processes, but many initial investments have failed to meet expectations. While language models (LLMs) can summarize information and suggest next steps, they lack the ability to act on that information without connections to key business systems. An LLM processing a supplier invoice might accurately summarize it, but it cannot match it to a purchase order (PO), determine payment status, or verify the correct general ledger (GL) account. To enable AI to actively participate in business processes, it requires access to the systems where relevant context resides. The Model Context Protocol (MCP) offers a standardized infrastructure for secure connections between AI applications and external data sources. By providing a common way for AI to access business systems, MCP simplifies integration, reduces complexity, and enhances AI's decision-making capabilities. With the right connections, AI agents can retrieve relevant data, make informed decisions, and take appropriate actions within defined business boundaries. This bidirectional connection allows AI to send information back to core systems, such as updating records, routing tasks for approval, or triggering subsequent steps. However, AI agents should not be granted unrestricted autonomy. The level of autonomy required varies depending on the task and implementation. As Forrester Principal Analyst Meng Liu noted, \"agentic branding is widespread, but real autonomy is not.\" That said, even limited autonomy can have a significant impact. In accounts payable (AP), a practical example of connected AI, AI agents can verify invoice details against purchase orders, delivery records, contracts, and GL accounts. This enables them to identify discrepancies, such as incorrect item prices, missing deliveries, or pricing deviations, and take appropriate action without human intervention. For instance, if an invoice's item prices don't match contract terms or delivery quantities, the AI agent can investigate further, request a vendor credit, route the discrepancy to the appropriate employee with recommended actions and supporting evidence, or approve the invoice if everything checks out. This same concept applies to other AP decisions, such as identifying potential duplicate invoices based on vendor history data or determining the correct GL codes using historical coding and approval history. The key metric for evaluating AI autonomy in AP is touchless processing, where an invoice is captured, validated, coded, approved, and paid without human intervention. Achieving touchless processing is a concrete way to gauge AI's effectiveness and mitigate potential issues like manual rework, overpayments, missed discounts, and inaccurate accounting records. While AI's success in other areas may be harder to measure, its impact on business processes can be significant, even without complete autonomy.",
  "summary": "If AI is given a more active role in business processes, this could increase efficiency and lower costs.",
  "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."
}