Quality assurance · Production

Pure Storage uses Augment to boost developer efficiency across a massive C++ codebase

The problem

Pure Storage's engineering teams struggled with a massive multi-language C++ codebase — debugging memory leaks and segmentation faults, navigating unfamiliar code sections, and accelerating onboarding ahead of an impending team merger. GitHub Copilot proved insufficient for these needs.

First attempt

GitHub Copilot, their initial AI coding assistant, was insufficient — its suggestions were context-unaware and unprofessional, and it could not handle the complexity of navigating the large C++ codebase.

Workflow diagram · grounded in source
1
Engineering challenge triggers AI use
trigger
“Facing challenges with a multi-million-line C++ codebase, Pure Storage leveraged Augment to enhance developer efficiency, improve onboarding, and streamline debugging”
2
Augment provides deep C++ context
ai_action
“Augment provided deeper context on templates, memory management, and complex class dependencies”
3
AI-assisted debugging
ai_action
“AI-powered insights helped engineers quickly identify and resolve runtime issues, memory leaks, and logical errors, significantly reducing debugging time”
4
Code navigation and Q&A
ai_action
“I could ask questions like, 'What does this do? How is it used? Why do we do it this way?' and get answers pointing me in the right direction”
5
Completions accepted into codebase
output
“Over 130,000 completions accepted into the codebase, with 77% of chat-driven suggestions implemented”
Reported outcome

Pure Storage accepted over 130,000 AI completions into the codebase with 77% of chat-driven suggestions implemented, reduced onboarding time from months to weeks, and significantly accelerated debugging and code navigation across engineering teams.

Reported metrics
C++ codebase sizeover 2.1 million lines of C++
Completions accepted into codebase130,000
Chat-driven suggestions implemented77%
Onboarding timereducing onboarding time from months to weeks
Show all 5 reported metrics
C++ codebase sizeover 2.1 million lines of C++
completions accepted into codebase130,000
chat-driven suggestions implemented77%
onboarding timereducing onboarding time from months to weeks
debugging timesignificantly reducing debugging time
Reported stack
AugmentGitHub Copilot
Source
https://www.augmentcode.com/customers/pure-storage
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Pure Storage accepted over 130,000 AI completions into the codebase with 77% of chat-driven suggestions implemented, reduced onboarding time from months to weeks, and significantly accelerated debugging and code navig…

What tools did this team use?

Augment, GitHub Copilot.

What results were reported?

C++ codebase size: over 2.1 million lines of C++; Completions accepted into codebase: 130,000; Chat-driven suggestions implemented: 77%; Onboarding time: reducing onboarding time from months to weeks (source-reported, not independently verified).

What failed first in this deployment?

GitHub Copilot, their initial AI coding assistant, was insufficient — its suggestions were context-unaware and unprofessional, and it could not handle the complexity of navigating the large C++ codebase.

How is this quality assurance AI workflow structured?

Engineering challenge triggers AI use → Augment provides deep C++ context → AI-assisted debugging → Code navigation and Q&A → Completions accepted into codebase.

Related quality assurance cases