Case Studies
/
Insurance
/

Phoenix reinvented policy servicing across every channel. In weeks.

Industry
Insurance
Company Size
Enterprise
About the company
Phoenix is one of Israel's largest multi-line insurance and financial services corporations. Managing over $200 billion in assets, it operates across a broad range of businesses, including asset management growth engines.

See what governed AI could do for your policy servicing

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Share

A premier insurance and finance giant already knew where they wanted to go. Together with their team, we got there fast, standing up governed AI for insurance operations that handles policy servicing across text, web, app, and voice. It answers the questions customers ask, takes the actions they need, and turns that into a distinct competitive advantage.

Phoenix was already ahead

Phoenix is one of EMEA's largest insurance and financial groups, with 77 years in the market, roughly $173B in assets under management, a $16B market cap, and about 4,000 people. Its book runs the full length of a customer's financial life, from life, health, long-term care, and travel, through motor, home, business, and liability, into pension, provident, education funds, savings and investment products, and on to credit and brokerage. This is a company that has earned its customers' trust over decades, and its leadership had already reached a clear conclusion. The value of AI shows up in the day-to-day operational workflows that customers feel firsthand.

That ambition met the reality of scale. Phoenix fields hundreds of thousands of customer interactions across text, web, app, and phone, with text-based channels carrying a large share of the total, week in and week out. Its existing tools were not built for that world. A legacy master bot and a DTMF phone menu routed people by rigid, pre-set rules, so customers were misrouted, staff absorbed the overflow, and self-service never quite rose to what modern policyholders expect. Phoenix's ambition was clear; their technology just hadn’t caught up yet. 

The impact on both sides of the business

Within weeks, the assistant went from a gated release to running across every channel in production, inside Phoenix's own cloud and through the company's full security and regulatory review.  Once live, it lowered the cost to serve in policy servicing, cut misrouted calls with intent-based routing, and lifted conversion on online applications. Customer satisfaction rose exponentially as automation grew, instead of being quietly traded away for cost savings, which is usually the tradeoff companies expect to make. These results prompted Phoenix to expand the same platform into claims, internal operations, and voice.

What we built together

Working shoulder to shoulder with Phoenix's team, we put governed AI agents across web, app, text, and voice, so a customer gets the same intelligent help no matter where they start. It handles policy servicing from beginning to end, answering policy, claims, and product questions using the company's own knowledge, the customer's own data, and the public marketing site, so the answers are accurate, personal, and grounded in Phoenix's policies rather than generic guesses. It also runs the sensitive flows that most companies are afraid to automate, including policy lookup, quotes, and claim status, each one wrapped in strict access controls and a complete audit trail on every interaction.

Notch's 5 layers of guardrails in AI agents

From there, we replaced the old master bot and the IVR outright. In place of a rigid phone menu, intent-based routing now understands what a customer is actually calling about and sends them to the right place on the first try. The entire platform runs inside Phoenix's own cloud in a hybrid architecture, and it cleared every security, regulatory, and legal review the company required before a single customer ever touched it. Governance was not something we bolted on at the end. It was the condition for going live at all, and Phoenix's team held that line with us the whole way.

Moving that fast is possible because of ADAM, the part of our platform that watches every deployment in production, finds the gaps, and updates workflows automatically. Each new workflow trains and reaches go-live faster than the one before it, which is how a carrier expands AI safely without slowing down.

And it kept going

Once the platform proved itself in production and the governance held up under regulatory scrutiny, Phoenix kept expanding it. The same trusted foundation now extends beyond that first support workflow into voice and IVR, claims, and internal operations, with each new tier bringing its own efficiency and savings on top of what came before. One workflow proven, then the next, then the next, instead of a single high-risk rebuild that a regulated insurer could never safely attempt.

"You need to be able to explain every decision it makes, keep it under control, and trust that it's consistently accurate. Notch has created a system that does just that."

Gil Tamir, Deputy CEO & Director of Innovation and Technology, The Phoenix

Phoenix set the bar. We helped them clear it, fast and together.
See what your first workflow could be. Book a demo.