Allianz Partners Cuts 1,800 Jobs and Exposes Where Insurance AI Truly Delivers
The travel assistance subsidiary confirmed the cuts via CEO Tomas Kunzmann on July 8. Comparing Germany, Japan, the U.S., and Brazil, the sector begins to mark where AI moves from the spreadsheet to the P&L.
On the afternoon of July 8, Tomas Kunzmann, CEO of Allianz Partners, confirmed what industry analysts had expected since the release of the latest Evident AI Index for Insurance. Allianz's travel assistance and insurance division will cut up to 1,800 jobs over the next 12 to 18 months, with most of the reductions occurring in European call centers. This represents a net reduction from a workforce of 22,600 employees, of which about 14,000 were handling claims by phone. Kunzmann's justification to Bloomberg was straightforward: automation of customer service and triage via AI.
Weeks earlier, Allianz had been ranked first among 30 global insurers in the Evident AI Index 2026 for the sector, ahead of AXA, with an AI talent pool 28% larger than the second place and over 900 mapped use cases. The distance between the top ranking and the announcement of the cuts is not coincidental, and it is what makes this round more instructive than usual: for the first time, an insurer is able to display an ordered correspondence between where it invested in AI and where it is pruning headcount.
Where AI is Actually Cutting
The emerging pattern is established along two axes. The first is high volume and low complexity. Travel claims, baggage theft, flight cancellations, 24-hour assistance: all cases where the script is short, documentation is standardized, and the average compensation amount does not justify human intervention on first contact. This is where Allianz operates Nemo, a system that runs triage and automatic disposition, and it is the same category that Travelers in the U.S. has been covering with a voice assistant for claims built on OpenAI models.
The second axis is auto claims, particularly with embedded telematics. The Japanese company Tokio Marine has already reduced the average auto claim closing time to 4.3 days with an AI-assisted portal, and Bradesco Seguros in Brazil has been operating automated total loss analysis in affiliated workshops since 2023, removing steps in the inspection flow. The German company ERGO, the primary subsidiary of Munich Re, announced cuts of about 1,000 positions in Germany attributed in part to the advancement of the same class of automation.
Where AI Remains Outside
The counterpoint comes from within the industry itself. A survey by McKinsey cited by actuaries shows that the adoption of AI agents in insurance is scaling quickly in risk, legal, and compliance (16%) and in knowledge management (16%), but remains between 0% and 2% in product development and strategy definition, which are exactly the areas where actuaries and underwriters concentrate their judgment. As noted by Roots.ai in an analysis of coverage response workflows, the error pattern of general-purpose models still generates confidently incorrect outputs, which in insurance regulation translates into immediate operational and reputational exposure.
The regulatory signal reinforces this boundary. Twenty-four U.S. states have already adopted the NAIC Model Bulletin on the use of AI systems by insurers, and an assessment tool is being piloted in another 12 by September. At the same time, AIG and W.R. Berkley are seeking regulatory approval to exclude liability for AI from standard corporate policies, a move opposite the internal enthusiasm of insurers that are automating internal processes.
What the Right Comparison Shows
The mistake in the public debate is to compare an insurer to an AI lab. An insurer is a portfolio of business lines with distinct regulatory sensitivities: travel and assistance have become operational commodities, low-value auto claims are on their way, but reinsurance, life with complex actuarial components, and corporate liability coverage remain human for reasons beyond nostalgia. The calculation that makes sense is not how many jobs AI eliminates in total, but in which line it moves from the spreadsheet to the P&L.
For Allianz Partners in Europe, this calculation weighed toward automation because the division thrives on repetitive volume. For ERGO in Germany or Bradesco Seguros in Brazil, the calculation is similar in mass auto. For Tokio Marine in Japan, the gain appeared as a reduction in closing window and not as mass layoffs, because the Japanese cost structure penalizes staff cuts and favors reassignment. This separates those making real transformation with AI from those merely counting cuts: the metric that matters is where the model entered as part of the product, not where the call center shrank as a consequence.