Cursor × FedEx
Shared securely for FedEx leadership
Incorrect password. Try again.
For FedEx · Prepared for Jim Connolly · June 2026

Accelerate FedEx application development without disrupting critical operations.

Cursor helps your teams pull the business rules out of legacy Oracle PL/SQL and rebuild them in Java, then test and certify the result, so FedEx can be off Oracle by 2029 without losing decades of operational logic.

See the use cases Model the ROI
Who is Cursor

The AI platform professional engineering teams build on.

Cursor brings AI agents, Tab, Cloud Agents, and automated review together in one platform, so engineering teams build software with AI across the full lifecycle, from plan to deploy. It is also a general-purpose agent, not just for the SDLC. With an SDK and deep integrations, you can point Cursor at almost any workflow and automate it, so the same platform your engineers use also powers co-work across the org.

70%+
of the Fortune 500 use Cursor
50,000+
enterprises build on Cursor
100M+
lines of enterprise code written per day
Trusted by teams at
NVIDIAStripeSalesforcePayPalSamsungDropboxCoinbaseAmplitude
Analyst recognition

A Leader in the 2026 Gartner Magic Quadrant.

Gartner Magic Quadrant for Enterprise AI Coding Agents, April 2026, showing Cursor positioned in the Leaders quadrant with the highest completeness of vision.

Cursor was named a Leader in the inaugural Gartner Magic Quadrant for Enterprise AI Coding Agents, with the highest completeness of vision of all vendors evaluated.

Read the report
Source: Gartner, Magic Quadrant for Enterprise AI Coding Agents, April 2026. Gartner and Magic Quadrant are registered trademarks of Gartner, Inc. and/or its affiliates and are used herein with permission. Gartner does not endorse any vendor, product, or service depicted.
How we differentiate

Three things that set Cursor apart.

Model flexibility

Use any AI model, no lock-in

Anthropic, OpenAI, Google, or Cursor's own model. Pick the best one for each job and switch whenever a better one ships. You never bet the whole program on a single vendor.

Purpose-built platform

Built for real engineering work

Cursor understands your whole codebase and works across plan, write, review, test, and deploy. It is a platform for the full job, not autocomplete in a single file.

Enterprise grade

Safe to roll out at scale

Privacy Mode, SOC 2, SSO, and admin controls over spend and model access come standard, so you scale across teams without losing control.

Why this matters to FedEx

The three whys.

1 · Why do anything

The legacy estate is slowing you down

~500,000 lines of Oracle PL/SQL in one monolith, roughly half of engineering capacity on keep-the-lights-on, and standards enforced by hand across teams. Delivery is slow and modernization is hard to execute.

2 · Why Cursor

The one platform that fits the job

Token-efficient and model-agnostic, it understands the whole codebase, enforces your standards with Rules and Skills, and runs agents across the migration. It works inside your GitHub workflow instead of replacing it.

3 · Why now

The real clock is tighter than 2029

Off Oracle by 2029, but with limited deployment windows rollout realistically starts in 2027. WebLogic and Maximo modernization run in parallel, so the sooner you prove the model, the safer the deadline.

Value framework
Cursor
Current → Desired state
FedEx × Cursor Transformation Map
PL/SQL locked in one monolith, 2029 Oracle clock running→Modular Java services, off Oracle ahead of schedule
Current State→
  • ~1,000 PL/SQL packages and 500 to 600K lines of code in one large GitHub repo.
  • Roughly half of engineering capacity spent on keep-the-lights-on work.
  • Coding standards, naming, and class structure enforced by hand across teams.
Negative Business Implications→
  • A large legacy surface slows delivery and makes modernization hard to execute fast.
  • The monolith is hard to decompose cleanly into modular apps and repos.
  • Manual prep and manual standards add overhead and risk inconsistent, hard-to-support code.
  • A tight timeline across PL/SQL, WebLogic, and Maximo raises the risk of delays.
↓ Required capabilities ↓
Required Capabilities
1Enterprise controls: Privacy Mode, SSO, model and spend governance
2A token-efficient platform, so developers actually use it and the enterprise can scale
3A model-agnostic platform, every frontier model in one place, no lock-in
4A strong harness that understands the whole codebase and gets more from every model
5Rules and skills that enforce FedEx standards across every team
6Automated code review on every change
7Cloud Agents for parallel, triggered migrations
8Plan each migration before writing, so no spaghetti code
↓ Desired state ↓
→Desired State
  • Modular Java and Spring Boot apps built from the business logic, not a line-by-line port.
  • A separate repo per app, giving cleaner ownership and parallel development.
  • AI-assisted modernization that fits your existing GitHub workflow.
→Positive Business Outcomes
  • Off Oracle ahead of the 2029 deadline, with rollout able to start sooner.
  • Faster progress across PL/SQL, WebLogic, and Maximo at the same time.
  • Maintainable, standard-consistent code your teams can support in production.
  • Engineering capacity freed from keep-the-lights-on for higher-value work.
Cursor Prepared for FedEx · Confidential
High-value use cases

Where Cursor earns its keep on day one.

Each use case is paired with a published result from a comparable enterprise engineering team.

Oracle exit

~500k lines of PL/SQL to Java

Rebuild roughly 500,000 lines of Oracle PL/SQL into modern Java applications, driven by extracted business rules and dossier specs, on track to be off Oracle by 2029.

Amplitude ships 3x more production code with Cursor.
Business rule extraction

Preserve the logic, not just the code

Cursor reads the existing PL/SQL and surfaces the business rules buried inside it, using your dossier specs as input, so the logic that runs operations is captured before anything gets rebuilt.

Dropbox indexed over 550,000 files to build an AI-native SDLC.
Rebuild in Java

A rebuild, not a line-by-line port

From the extracted rules and your specs, Cursor helps generate the new Java applications, so a smaller team can tweak, test, certify, and deploy on the path off Oracle.

National Australia Bank accelerated legacy migrations with Cursor.
Cloud & platform migration

Move migration work faster

Cursor accelerates code changes, dependency mapping, documentation, and test creation for migration initiatives such as Azure or hybrid cloud modernization.

Salesforce accelerated velocity 30%+ and shipped higher-quality code.
Review & test

Unblock where the train stalls

Review and testing are the bottlenecks slowing the release train today. Cursor generates unit, regression, and validation coverage around critical logic, and Bugbot reviews every change before it ships.

PlanetScale protects production reliability with Bugbot.
Onboarding & knowledge transfer

Ramp new and rotating engineers faster

Cursor helps new or rotating engineers understand unfamiliar applications faster, reducing ramp time and the risk that comes with handoffs across a large, distributed org.

Stripe rolled out a consistent Cursor experience for 3,000 engineers.
Repo strategy

One monorepo to clean Java repos

Your PL/SQL sits in one GitHub repo. Cursor runs the GitHub CLI to spin up the new Spring Boot repos, move the code across, and scaffold each service, all captured as a reusable skill so the split is repeatable, not manual.

Encoded as a reusable Cursor skill, run it for every app.
The principle

Modernization, not conversion.

This is not a line-by-line port of PL/SQL into Java. It is extracting the business rules, rebuilding clean Java apps from your dossier specs, and moving toward true agentic development: people own the prompts and specs, AI does the writing, and a smaller team reviews, tests, certifies, and deploys. The bigger shift is building the higher-level agents that run the factory itself, so the migration keeps producing and stays maintained going forward. It can even reach past engineering, so operators answer their own operational questions with AI instead of waiting on a new app for every use case.

Proven at scale

Shipped results from enterprise engineering teams.

How engineering leaders at Intuit, DoorDash, and Atlassian are adopting AI coding.

3x
more code committed across 30,000 developers
NVIDIA
90%
faster from idea to production
Coinbase
2x
PR throughput with Cloud Agents
Faire
Trusted by engineering teams at
NVIDIAStripeSalesforcePayPalSamsungDropbox
And across transportation & logistics
MaerskNorfolk SouthernRXOHertzAir CanadaSIXT
Business value

What it adds up to.

The outcomes a director over four operational teams can take to the rest of the org.

Compress the rewrite

A migration-factory model turns a multi-year PL/SQL rewrite into a repeatable pipeline.

Free the 50%+ on keep-the-lights-on

Hand routine maintenance to agents and put scarce capacity back on the migration.

Unblock review and test

Move faster where the release train actually stalls today.

Avoid runaway cloud cost

Rebuild clean instead of lifting brittle PL/SQL architecture into the cloud unchanged.

Hit 2029 with buffer

Start rollout in 2027 and finish early, inside FedEx's limited deployment windows.

Less manual admin

Cut the documentation and process overhead weighing on developers.

ROI projection

Model it for your teams.

An illustrative, attribution-based estimate. Move the sliders to match your rollout. Defaults reflect a four-team starting point, not a full-company deployment.

Projected annual value
$2.7M
Gross productivity value$4.5M
Estimated Cursor cost$146K
Net annual value$2.7M
Return on investment18×
Illustrative model only. Gross value = developers × fully-loaded cost × velocity uplift. Net applies an attribution discount. Cursor cost basis is an estimate of roughly $122 per developer per month ($40 seat + $50 usage + $32 review). Actual results vary by team and codebase.
Pilot recommendation

A focused 30 to 45 day pilot with one application team.

Scope it to a single team, a single codebase, and one measurable outcome. We cover the token costs for the pilot. Choose the starting point that maps to your biggest pain.

Option 1

PL/SQL to Java rebuild

Take one Oracle PL/SQL module from the 2029 exit list and measure how fast a small team can extract its rules, stand up the Java replacement, and pull the 2028 rollout forward.

Option 2

Test coverage improvement

Generate unit, regression, and validation coverage around critical operational logic and measure the lift.

Option 3

Developer onboarding acceleration

Put new or rotating engineers on an unfamiliar app and measure the reduction in ramp time.

Option 4

Business-rule extraction

Point Cursor at a dense PL/SQL module and measure how completely it captures the business rules from the code and your specs.

How it works

From handoffs to an AI software factory.

Today, work passes from person to person at every phase, and the handoffs are where time is lost. Cursor turns that line into a loop of AI agents, with your engineers directing instead of doing each step.

Today: work is handed off at every phase Plan Design Write Review Test Deploy handoffs lose time With Cursor: one team directs a factory of agents P D W R T D AI Software Factory idea in shipped
Agents run the full development lifecycle, with engineers directing instead of doing each step.
The migration flow

From PL/SQL to running Java services, inside Cursor.

Legacy apps and dossier files in on the left. Cursor does the four steps in the middle. New Java services ship to GCP or on-prem on the right, and the rules and skills repeat the pattern for every app.

Old PL/SQL apps + dossier files Cursor 1. Understand the rules 2. Rebuild in Java 3. Test and review 4. Prove it matches New Java service GCP On-prem Rules + Skills: repeat for every app
Repo strategy

One PL/SQL repo in. Clean Java repos out.

Today the PL/SQL lives in one large GitHub repo. As you modernize, each app becomes its own Spring Boot repo. Cursor runs the GitHub CLI to create the repos, move the code, and scaffold each service, captured as a reusable skill so it is encoded once and repeated for every app.

PL/SQL monorepo one big GitHub repo Cursor skill runs the GitHub CLI: create repos, move code New Java repo Spring Boot service New Java repo Spring Boot service New Java repo Spring Boot service
One repo per app, created and wired up by a reusable skill, not by hand.
The journey

The AI maturity curve.

Every engineering org ascends this curve. FedEx sits between stages two and three today. The goal is to move up it safely and quickly, and the agentic stages in the middle are exactly where Cursor operates.

FedEx today
1
Manual coding
AI not in the loop
2
Assisted coding
AI as assistant
Cursor Tab
3
Prompted blocks
AI as pair
Cursor Agent
4
Guided feature gen
AI as junior eng
Cloud Agents
5
Autonomous dev
AI as team
Agentic loops
6
Intent-oriented
AI as cofounder
Factory
What blocks the ascent
1

Developer trust

Adoption sticks when developers choose the tool, not when it is mandated.

2

Model lock-in risk

The best model changes every few months. Single-vendor bets lose.

3

Developer relapse

Without governance, engineers revert to old habits. Adoption needs structure.

4

Traceability and audit

AI-generated code needs an audit trail, required for compliance.

5

Security and controls

Must fit enterprise security, infrastructure, and data residency requirements.

FedEx is between stages two and three. Cursor enables stages 2 through 5 today, and the blockers on the right are exactly what it removes, so the climb stays safe and fast.
Your pilot

Your 30-day legacy modernization pilot

A focused, low-risk pilot with one application team. Cursor covers the token costs. Five phases, left to right.

Download the one-page charter (PDF)

Setup

Day 0–3

Lock one team on a real Oracle PL/SQL module from the 2029 exit list (8–15 engineers). Privacy Mode and SSO on. Tokens covered by Cursor. Success criteria signed off.

Kickoff

Week 1

Jason Wiker, our field engineer, leads a tailored engineering session and a casual demo of a few sample migrations. Pick one real legacy module plus one live defect or regulatory change.

Run

Weeks 1–4
  • Comprehend. Cursor reads the PL/SQL and extracts and explains the business rules.
  • Change. The agent drafts the Java rebuild from your specs and opens a PR.
  • Review. Automated code and security review with Bugbot.
Mid-pilot working session captures before and after.

Day 30 results

Outcomes

Business rules from one PL/SQL module captured. One Java application stood up through the agent and review workflow. Engineer confidence on the rebuild path up.

Expand

Scale

Roll to more app-dev teams, move into corporate approval and procurement, and scale across Ground and Services.

Plan & owners

Mutual action plan.

The sequence from here to a proven pilot, and who owns each step. Steps run in parallel where they can.

1
NDA in place
FedEx sends its MNDA/CDA; Cursor signs the same day.
FedEx · JimCursor · Joe
This week
In progress
2
InfoSec first request
Bryan compiles the first-request questions; Cursor answers them.
FedEx · BryanCursor · Jason
This week
In progress
3
InfoSec review complete
Required before any FedEx code is used in the pilot.
FedEx · InfoSecCursor · Jason
Before code goes in
Next
4
Hands-on now, personal accounts
Bryan and Greg sign up with a personal email; Joe provisions token credits; Jason runs a working session. Personal or side projects only until the pilot is approved.
FedEx · Bryan & GregCursor · Joe + Jason
This week
In progress
5
Procurement path chosen
Decide direct vs buying through AHEAD, whichever is the fastest route.
FedEx · JimAHEAD · MikeCursor · Joe
In parallel
Next
6
Pilot setup
Provision accounts, SSO, and admin controls in a short session.
Cursor · JasonFedEx · Bryan
At kickoff
Planned
7
30-day pilot
One PL/SQL module rebuilt in Java from the dossier files. Cursor covers the token costs.
FedEx · Bryan's teamCursor · Jason
30 days
Planned
8
Warm intro to Mario Jones (SVP)
Jim introduces us to his SVP, Mario Jones, to sponsor the broader modernization program.
FedEx · JimCursor · Joe
This week
In progress
9
Executive alignment
FedEx and Cursor leadership connect to align on the broader program.
Cursor · JoeFedEx · Jim
Jul 8–9
In progress
10
Review and decision
Measure results against the success criteria, then move to contract.
FedEx · JimCursor · Joe
Post-pilot
Planned
Next step

Start small. Prove value. Scale what works.

Identify one team, one codebase, and one measurable outcome where Cursor can help FedEx validate productivity, modernization, and quality gains.

Start a conversation Download the pilot charter (PDF)
Prepared by Joe Masello · joe.masello@cursor.com · June 2026