BT · Amazon
200,000 — Lines of code written
Independently reportedTelecom · Code generation · LLM · In production
Code generation
allpay · Microsoft
Vendor-reportedBanking & financeUnited KingdomIn productionMid-marketLLM
Vendor-reported. The customer is named and the numbers are quoted from the source page, but the account comes from the vendor. No independent confirmation.
Reported by microsoft.com (vendor self-report). Checked against the source page on 2026-09-06. 4 of 4 figures below appear on that page word for word.
The problem
Existing IT infrastructure would not support future growth plans, requiring acceleration of software development, improvement of code quality, and better understanding of legacy code.
What was deployed
- Vendor
- Microsoft
- Products
- GitHub Copilot, Microsoft Copilot
- Technique
- LLM
- Build or buy
- Bought off the shelf
- Deployment
- SaaS
- Data used
- codebase
- Scale
- 50-999 employees (per Microsoft's own case metadata); development organized into 12-14 squads
What changed
Each row is quoted from the source. Figures we could not find on the page in those words are marked — they are kept, not deleted, so you can judge them.
increased by 10%developer productivity
“allpay estimates that, overall, the team’s use of GitHub Copilot has increased developer productivity by 10%.”
Quoted word for word from the source · round percentage, no baseline given
80%time saving for some use cases
“delivering as much as an 80% time saving.”
Quoted word for word from the source · round percentage, no baseline given
increased by 25%delivery volume into production
“In combination with some organisational changes, the adoption of GitHub Copilot has driven up delivery volume into production by 25%.”
Quoted word for word from the source · round percentage, no baseline given
more than 900 delivered in nine monthsannual releases to productionfrom 900 delivered in the whole of last yearnine months
“In terms of releases out the door, in the nine months of this year to date, we have already delivered more than the 900 delivered in the whole of last year.”
Quoted word for word from the source
droppedtime spent on rework and defect fixing
“The amount of time we are spending on rework and defect fixing has dropped”
Quoted word for word from the source
fewerincidents
“We’re getting fewer incidents.”
Quoted word for word from the source
“At the same time as getting codes out the door faster, GitHub Copilot is helping us to raise the quality of code, testing, and applications.”
Difficulties and limits
- Consistency across 12 or 14 development squads is hard to manage without dedicated cross-team effort
- More experienced developers don't always accept Copilot's suggestions, while junior team members find them especially useful — acceptance rates differ by seniority
Sources
- microsoft.com — vendor self-report
Similar deployments
Finger Lakes Daily News · TOTIB
99 — Mobile PageSpeed
Vendor-reportedMedia & entertainment · Code generation · LLM · In production · 2026
Shriram Finance
less than 3–4 hours — Onboarding Turn-Around-Time (TAT)
Independently reportedBanking & finance · Process automation · Classical ML · In production
ANZ Bank · NVIDIA
0.82 — Gini coefficient for assessing risk
Independently reportedBanking & finance · Other · Classical ML · Pilot
Radiant Credit Union · interface.ai
60% — Asset growth absorbed without contact center team expansion
Vendor-reportedBanking & finance · Customer support · LLM · Scaled · 2022
DNB · boost.ai
2,500 — Relevant topics covered
Vendor-reportedBanking & finance · Customer support · NLP · Scaled · 2018