Tutorial: use Apex AI to generate Apex test classes in Salesforce. Connect code, generate tests, review assertions, run tests, iterate, and ship.
Shipping reliable Salesforce changes depends on high-quality Apex test classes-yet writing them can be time-consuming, repetitive, and easy to under-scope. In this tutorial, you’ll learn how to use Apex AI to generate Apex test classes that are tailored to your org’s code and ready for review in a modern Salesforce engineering workflow.
## What you’ll accomplish
By the end of this guide, you’ll be able to:
- Generate an Apex test class from an Apex class or trigger
- Validate coverage and assertions without sacrificing code quality
- Iterate quickly when logic changes (or when a test fails)
## Prerequisites
Before you start, make sure you have:
- Access to Apex AI (your team workspace and a project)
- Your Salesforce code available (from a repo or local workspace)
- A target Apex class or trigger you want to cover (e.g., a service class, handler, or trigger)
## Step 1: Connect your source code to Apex AI
1. Open Apex AI and create/select a project.
2. Add your codebase (commonly via a repository integration or by uploading the relevant Apex files).
3. Confirm Apex AI can see the dependencies of the target file (helper classes, selectors, domain classes, etc.).
**Tip:** Better inputs produce better tests. Include related classes that influence branching logic-especially validation rules, helper utilities, and any platform event or queueable orchestration.
## Step 2: Choose the Apex class or trigger and generate tests
1. In your project, navigate to the Apex file you want to test.
2. Select **Generate Test Class**.
3. Configure generation options:
- **Target type:** Apex class, trigger, or handler
- **Testing style:** positive/negative paths, bulk scenarios, exception cases
- **Data strategy:** use `@testSetup`, factory methods, or inline test data
4. Run generation.
Apex AI will produce a test class scaffolded with:
- `@isTest` annotation and named test methods
- Structured setup and execution sections
- Assertions designed to verify outcomes (not just increase coverage)
## Step 3: Review the generated test class like production code
Treat generated tests as a first draft. Review for:
- **Meaningful assertions:** Do tests validate outputs, DML results, state transitions, and thrown exceptions?
- **Bulk behavior:** Are there tests inserting/updating lists (e.g., 200 records) to catch governor-limit issues?
- **Edge cases:** Null inputs, missing related records, invalid statuses, empty collections
- **Isolation:** No reliance on org data; confirm `SeeAllData=false`
If your team uses a test data factory pattern, align the generated test to your conventions (naming, builders, and reusable setup).
## Step 4: Run tests and iterate using Apex AI
1. Run your test suite in your preferred toolchain (SFDX, Salesforce UI, or CI).
2. If a test fails:
- Copy the failure output (exception message, stack trace, failing assertion).
- In Apex AI, choose **Refine** or **Fix Test** and provide the failure context.
- Regenerate the test method(s) or adjust only the failing sections.
This loop is where Apex AI shines: you can converge on stable, high-signal tests quickly as business logic evolves.
## Step 5: Commit, open a PR, and keep tests maintainable
Once tests pass:
- Ensure the test class naming matches your standards (e.g., `MyService_Test`).
- Add comments only where intent isn’t obvious.
- Keep methods focused (one behavior per test) and avoid brittle assertions.
- Commit and open a pull request so reviewers can validate coverage and intent.
## Conclusion
Apex AI helps Salesforce engineering teams generate Apex test classes faster while keeping quality in check through review and iteration. If you want to reduce time spent on repetitive test writing and ship changes with more confidence, start by generating tests for one high-impact class today.