Epiq is a global leader in the legal services industry, taking on large-scale, complex work for corporate counsel, law firms, and business professionals across litigation, investigations, eDiscovery, and compliance. Its flagship platform, Epiq Discover™, runs the full eDiscovery lifecycle from processing through review to production, built by engineering teams across the US, Poland, and India.
Epiq experts Trevor Best, Senior Director of Product Development, and Alex Kapadia, Vice President of Application Support, oversee engineering teams that span multiple countries. Both were grappling with the same challenge: AI was enabling them to ship software faster, and they needed to be faster at responding to new bottlenecks.
AI sped up coding and moved the bottleneck
Before LinearB, Epiq engineering data lived across GitHub Insights and Azure DevOps, with no easy way to aggregate it across pull requests or teams. As AI coding tools rolled out, that blind spot became a liability. Adoption was near total, to the point where Trevor treated writing code by hand as a competitive risk.
Yet all that extra AI-generated code wasn't translating into faster delivery. Cycle time held flat, and without a way to see across the pipeline, no one could say why. When Trevor's team dug into the data, the reason was clear. AI had made writing code fast, but each change still needed a human review that hadn't sped up to match.
Trevor Best
The symptoms showed up downstream. Pickup time climbed as the volume of AI-generated pull requests outpaced the humans available to review them, and changes routinely waited days before an engineer could get to them.
Epiq isn't alone in this, as their experience matches the broader industry trends found in LinearB’s 2026 Software Engineering Bechmarks Reoprt. 
Moving review upstream, and proving the shift with data
The answer for Epiq was to move review earlier in the pipeline. Rather than waiting for a pull request to reach a human, it ran AI review agents while code was still being written.
Trevor Best
LinearB gave Epiq the data to watch that shift happen in real time, including the point where AI reviews overtook human-only ones.
The same visibility now feeds a harder internal question. With AI able to deliver a fully tested, fully passing feature in a single pass, are the Epiq long-standing pull request (PR) size limits still the right guardrail?
Trevor Best
A redesign like this is driven by managers and directors, not just executives, so Epiq needed a tool built for them. When it weighed its options, LinearB pulled ahead of Jellyfish, which some of its leaders had used before. Jellyfish was strong for the boardroom, built for the reporting and finance conversation, but Epiq needed something its team leaders could act on directly.
Alex Kapadia
Epiq wanted to act on what the data showed while it still mattered, not read about it a quarter later.
Fixing the right problem, and finding value beyond it
The clearest gain was knowing where to look. Instead of treating slow cycle time as a coding problem, Epiq could see it was a review problem, and put its effort into the constraint that was actually holding delivery back rather than the one that was easiest to see. Mechanical checks now run earlier in the pipeline, so human reviewers spend their time on the judgment calls only a human can make.
The same data kept paying off in places the team hadn't set out to fix. Comparing AI adoption across the US, Indian, and Polish teams gave Epiq leaders visibility they'd never had into where AI was taking hold, and where it wasn't.
Alex Kapadia
And because LinearB shows who is actually writing code, Epiq could line that up against its accounting records and spot senior engineers whose new-feature work wasn't being logged as a capital investment. Alex ballparked the gap at around US$50,000 for one team over a few months.
The same visibility carried into how Epiq leaders run reviews, giving them a consistent, data-informed basis for evaluating their teams.
Rewriting the rules built for a pre-AI world
For Epiq, fixing the review bottleneck was only the start. It showed how much of their process had been built before AI and was overdue for a rethink. The team is working through that rulebook now, starting with the fixed PR size limits that no longer fit a world where AI can build and fully test a feature in one pass. With LinearB giving them a clear view of how work moves, Epiq can keep redesigning its process as fast as AI reshapes the work.
See how LinearB can help your engineering team today. Book a demo.
