Liberty Mutual told its engineers to try serverless first, then cloud-native, then containers. It built a library of CDK patterns in TypeScript, deployed 3,500 of them in a year, cut build times from a year to three months and runs a global ledger at $60 per million transactions. What made it work.
A written serverless-first policy: engineers build serverless first, cloud-native services second, and containers only when neither fits.
A software accelerator: reusable serverless patterns written in TypeScript with AWS CDK. More than 3,500 were deployed between 2019 and 2020.
Financial Central Services, started in 2019, consolidated lines of business worldwide into one general ledger, with the extract-transform-load process orchestrated by Step Functions. It processes about 100 million transactions a month at roughly $60 per million, and costs almost nothing when idle, which is most of the month.
Application build time fell from about a year on-premises to about three months. A four-person team built an equine and livestock insurance application in twelve weeks.
About half of workloads were on AWS by 2020.
Book a free 30-minute call. You'll leave knowing what we'd do first and roughly what it costs, whether or not you hire us.
The policy did more than the technology. A default that engineers must argue against changes architecture across hundreds of teams in a way that guidance never does.
The pattern library is the multiplier. A team that can start from a reviewed CDK construct for "API plus queue plus table" ships in weeks, and security and cost standards travel with the pattern.
Serverless fits batch-shaped finance work well. A ledger process that runs at month-end and sits idle the rest of the time is the textbook case for paying per use.
It is still a mix. Serverless-first is not serverless-only, and the policy says so.
Once a month: what changed in tech that month and what it means for you, from cloud and AI to security and the tools you build with. Every item links to its source. No sales pitch, unsubscribe in one click.
Book a free 30-minute call about it. Bring your setup and your bill; you'll leave knowing what we'd do first and roughly what it costs, whether or not you hire us.
Or keep reading
What Ten Enterprise Modernizations Have in Common: Home Depot, Lowe's, Capital One, Netflix, Amazon and More
Ten public modernization stories, side by side: where each started, how long it took, how much was rebuilt rather than moved, and what changed. Six patterns repeat in almost all of them, and each one applies to a company far smaller than these.
Read
How Amazon's Consumer Business Turned Off Its Last Oracle Database: 7,500 Databases, 75 PB, 100 Teams
In October 2019 Amazon's consumer business finished moving about 7,500 Oracle databases and 75 PB of data to DynamoDB, Aurora, RDS and Redshift, with little or no downtime. It cut costs by more than 60% and consumer-facing latency by 40%. How the work was split, why each workload went where it did, and what applies to a company with ten databases.
Read
As of October 25, 2026. Provided as is, for general information, and accurate to the best of our knowledge on the date shown. Not legal, compliance, tax, financial or other professional advice. Prices, limits and vendor services change; check current vendor documentation before relying on anything here. Terms of use.