← Research
Agents & workflows04 Nov 2026 · 2 min

AI agents in the enterprise: who scales, who fails

AI · not reviewedWritten with AI from the sources listed, not yet editorially reviewed.

Large companies are deploying agents more broadly, smaller ones barely — and analysts expect many projects to fail on cost and unclear value.

The essentials in five points

  1. According to McKinsey, 40% of large companies (USD 1 billion+ revenue) are scaling AI agents in 2026, up from 27% a year earlier; among smaller ones, the share is unchanged at 22%.
  2. Companies most often scale agents in IT, knowledge management and software engineering; 37% of respondents attribute at least some EBIT impact to AI overall.
  3. In Germany, 11% of companies that use or plan AI already deploy agents; 29% plan to (Bitkom, 603 companies, summer 2026).
  4. Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027 — due to rising costs, unclear business value and inadequate risk controls.
  5. Of thousands of vendors, only about 130 deliver genuine agentic capabilities, according to Gartner; many simply rebrand existing chatbots or RPA tools (“agent washing”).

What this means for businesses

Agents pay off where processes are clearly described, data is accessible and results are measurable — for example in IT service or knowledge search. Start with one process, one metric and clear approval limits, and check whether a vendor’s product actually acts on its own or is just a renamed chatbot.

Sources · 3

  1. McKinsey & Company · August 2026The state of AI in 2026: On the road to ROI
  2. Bitkom · 14 Sept 2026Erstmals nutzt die Mehrheit der Unternehmen KI
  3. SD Times (Gartner-Prognose) · 30 Jun 2025Gartner: More than 40% of agentic AI projects will be canceled in the next few years

Links open the original source in a new tab. Figures from company statements are labelled as vendor or company figures in the text.

More from this branch

All in Agents →
  1. Agents

    AI agents in 2026: where they actually work

    AI agents now do real work, mostly in coding and customer service, but they still fail often and are being adopted more slowly than promised.16 Oct 2026

The next OROBO Report comes to you.

Planned from December 2026, every Thursday. Free, unsubscribe in one click.

Subscribe ↓