Sabermine passes five million cases in under three years

Sabermine has processed more than five million cases for regulated firms, reaching the milestone in less than three years.

The Manchester software company was founded in January 2024 by engineers Jakub Lenski and Mitchel Shephard after an audit project for a law firm exposed a large volume of manual correspondence and document work in group litigation.

The assignment involved a 200,000-claim book that had been expected to take weeks to review by hand. Sabermine built a one-off automation workflow that completed the task in under three days, prompting the founders to turn the work into a company.

Since then, the business has grown to 32 employees, most of them engineers, and expanded from litigation and mass claims into insurance, pensions administration and credit. It recently reached £13 million in annual recurring revenue without external funding, paid customer acquisition or a dedicated sales team.

Early growth

Sabermine focuses on high-volume operational work in regulated settings, where businesses often still rely on staff to move information between inboxes, documents and internal systems. Its software sits on top of a firm's existing tools rather than replacing them, which the company argues shortens deployment times.

Lenski said speed of deployment has been central to the company's pitch to customers in sectors where technology projects can stall before reaching day-to-day operations. He linked that to wider doubts over whether artificial intelligence systems can be trusted in regulated environments without human review.

Human oversight remains part of the model, with staff approving work before actions are finalised. Sabermine says this approach has helped it gain adoption in areas where errors can carry legal or financial consequences.

Customers typically see the software go live on a production workflow in about two weeks. That is significantly faster than many automation projects in regulated industries, where long integration cycles can delay deployment and weaken the original business case.

"Most automation projects in regulated sectors take six months to go live," said Jakub Lenski, Co-Founder, Sabermine. "By then the workflow has moved on, the team has lost faith, and the original business case is unrecognisable. We built Sabermine so the platform is live on a real production workflow within two weeks - and the customer proves the savings on their own data before signing anything commercial."

The company says its system saves 96.3% of time per case, although it did not disclose the customer base behind that figure. The five million-case total suggests the software is now being used at meaningful scale in sectors that handle large volumes of claims, policies and accounts.

Different model

Sabermine's growth stands out in a UK technology market where many young artificial intelligence companies rely on venture capital to expand sales, hire staff and market products. The company has taken a different route, growing organically and largely through customer referrals and proven use cases.

That approach has also raised its profile beyond its home market. Sabermine was named the UK's Technology Startup of the Year and National Overall Startup of the Year at the UK Startup Awards.

Lenski, 25, previously worked as a Software Engineer at Arm and had started businesses while still studying before co-founding Sabermine. His account of the business suggests a narrow operational problem, rather than a broad attempt to build a general-purpose AI product, shaped the company's early direction.

The broader backdrop is a business market still trying to turn AI experimentation into dependable operating tools. Many companies have tested generative AI systems over the past two years, but a significant number have struggled to move projects from pilot stages into daily production, especially in heavily regulated fields.

Sabermine argues that its appeal lies less in the underlying models than in how the software is deployed around existing workflows and controls. By fitting into established systems rather than requiring customers to rebuild operations around a new platform, it aims to lower the risk of adoption.

"We didn't have a marketing function, a sales team or a pipeline," said Lenski. "What we had was a result a customer could verify on their own data in two weeks. In this market, that proves to be more persuasive than anything else."