Insurance has long carried a reputation for being slow, bureaucratic and paper-heavy. However, a quiet shift is underway. Intelligent AI agents are starting to transform the industry, promising faster claims, proactive support and personalised policies.
At Cypher 2025 in Bengaluru, Sayandeep Majumder, AVP of data science and AI engineering at AXA GBS, took the stage with a different perspective. He suggested that the future of insurance could resemble a James Bond movie, albeit without the Aston Martins crashing constantly.
While Bond may rely on gadgets and partners, Majumder’s focus was on the real “secret agents” quietly working inside AXA’s systems.
He pointed out that insurance does not need to remain reactive and cumbersome. Instead, with AI agents, it can become proactive, fast and surprisingly human-friendly.
From Claims Chaos to Connected Agents
To make the point, Majumder painted a Bond-style scene. “From minutes of the crash, as soon as Bond calls up Q and notifies him of the Aston Martin’s ill fate, our agents get into action,” he said.
Fraud checks, repair assessments, satellite data, and cost approvals could all be triggered instantly. “Bond can just walk away with his almost new Aston Martin while we pick up his bills yet again.”
He explained that the level of efficiency once required teams handling multiple screens and endless forms.
Now, AI agents can perform the same tasks four times faster, reducing the manual workload. The real innovation, however, lies in how these agents communicate with each other. Using structured message protocols, they coordinate tasks like specialised departments in a mission, reducing the need for constant human supervision.
Majumder pointed out that legacy IT systems often slowed down insurers.
AXA, like many global organisations, has accumulated a patchwork of old and modern platforms over the years.
Here, Model Context Protocols become useful. By allowing AI agents to integrate with existing tools, they enable new intelligence to operate without requiring a complete overhaul of existing technology.
The bigger ambition, though, is not speed alone. “It’s not enough for us to build solutions using individual GenAI agents. We are talking about a strategic transformation that changes how we approach product development and customer engagement at its heart,” Majumder emphasised.
Insurance products could become dynamic, shifting in response to customer needs, rather than being fixed contracts. Claims would not just be responses to accidents, but proactive moments of support.
Beyond the Hype, Towards Human-Centred AI
Despite the optimism, Majumder added a note of caution. AI headlines often promise either a dystopian or a utopian future.
“The media loves sensational headlines, right? 300 million jobs at risk, all doctors gone, 40% of knowledge workers gone. These clickbait stories do not capture the nuanced truth,” he reminded the audience.
At AXA, the approach starts with real problems and tests relentlessly. He shared an anecdote where a software tester ordered 1,100, then 9,999 beers from a bar to complete the test system.
However, when a real customer asks to be shown the restroom, the bar explodes. This highlights the need to design systems for real-world applications, not just lab scenarios.
He also noted the limits of guardrails. Citing research, Majumder noted that unsafe responses in AI systems can jump from “0.3% to 9.2%” even with simple changes. For insurers handling sensitive situations, this margin is critical.
Yet, Majumder did not present AI as a threat to human roles, citing examples from the company. “AI is not about replacing jobs. It’s about elevating them,” he said.
Operations staff now supervise and guide agents rather than being overwhelmed by manual screens. Developers delegate coding tasks to assistants but still review and refine. Testers, numbering in the tens of thousands across AXA, can generate test cases instantly and focus on quality assurance.
“Humans with AI are going to replace humans without AI,” he added.
He also explained that the company’s global capability centre in India has become the proving ground for this shift. Acting as a bridge between research in Paris and delivery teams worldwide, it pilots tools, evaluates platforms and decides where agents deliver genuine value.
What Majumder offered was not a silver bullet but a blueprint for gradual reinvention. AI agents, working quietly across underwriting, claims and service, may not have Bond’s glamour, but they could be the most important partners the industry has ever had.
“It is refreshingly simple. We want to make insurance better for all of our customers, but one agent at a time,” he concluded.
The post AXA is Trying to Revolutionise Insurance With One Agent at a Time appeared first on Analytics India Magazine.
