Cursor sped up development 10x and left a financial company with a million lines to review

AI-powered programming tools are dramatically increasing development speed, but simultaneously creating a new operational crisis: companies physically cannot keep up with checking and securing the code they themselves generate. One case in the financial sector showed how quickly the benefit can turn into overload.

Code volume explosion

After implementing Cursor, one financial services company almost instantly changed its development pace. Code output grew from 25,000 lines per month to 250,000 — effectively tenfold. In a short period, this amounted to about a million lines that needed to be reviewed, tested, and run through security processes. While generation moved forward, internal control mechanisms began to fall behind.

Such a jump looks like an efficiency win only at the first stage. When code appears faster than teams can read and understand it, acceleration begins to work against the product. The larger the volume, the higher the chance that errors, weak logic, and vulnerabilities that no one noticed in time will enter the system. AI removes some of the burden from developers, but doesn't eliminate the need for human oversight.

The verification bottleneck

According to Jonny Klippert, co-founder and CEO of StackHawk, the problem isn't just about the growth in lines of code, but the growth in associated risks. The faster a company produces changes, the more reviews, tests, and security checks it needs. If these processes don't scale alongside generation, a huge backlog of tasks emerges that slows releases and increases the likelihood of incidents.

"The huge volume of code being created and the growth in

vulnerabilities — this is something they can't handle."

• code review and architectural review
• automated and manual testing
• vulnerability scanning and bug fixes
• documentation updates and internal instruction changes
• alignment of changes between teams

If a company doesn't reshape these stages beforehand, AI assistants don't just deliver a speed boost, but also introduce a new layer of technical debt. Code accumulates faster than the organization can process it. As a result, the winner isn't the one generating more lines, but the one who can faster distinguish ready-made changes from dangerous and raw ones. For regulated industries, this is particularly painful, because a failure in review can result in both risk and fines.

Pressure on business

The story doesn't end with engineers. When product output accelerates, it automatically pulls other departments along: sales need to explain new features to customers faster, marketing needs to update communications, and support needs to handle new scenarios and errors. According to Klippert, it's precisely this chain reaction that creates "enormous stress" in companies, because the pace changes for everyone at once, not just the development team.

This is an important signal for businesses implementing AI coding as a way to simply do more for the same money. In practice, the problem shifts: previously the bottleneck was writing code, now it increasingly appears in review, deployment, and maintenance. If sales, onboarding, support, and security processes aren't ready for the new pace, the company gets not acceleration, but chaotic overload.

What it means

AI assistants like Cursor can indeed dramatically accelerate development, but by themselves don't make a team more mature. The main question now isn't how much code a model can produce, but whether a company can quickly review that volume, safely deploy it to production, and not break other business functions along the way.