Should you write tests manually before using AI to generate them?

Cover

Photo: StockSnap (CC0)

Spotlight

PAGE: https://alison.com/course/rest-api-manual-testing-for-beginners

PROMISE: Gains a clear baseline for manual API tests before automating or using AI.

WHO: Beginners new to testing who need a fast, structured foundation before moving to automation or AI.

WHYNOW: Setting a correct manual pattern early prevents AI from scaling broken or incomplete test logic.

DESC: This app development course explains client server architecture and how to use REST to manually test applications with tools like JSON, CRUD, Postman and cURL.

BEFORE: Relies on guesswork when deciding which test cases to write first

BEFORE: Lacks a concrete method to map business actions to API calls

BEFORE: Has no structured approach to separating manual setup from automated generation

AFTER: Uses the course definitions to scope which tests matter for the project

AFTER: Applies the CRUD-to-HTTP mapping to write the first few tests manually

AFTER: Selects Postman or cURL to execute those manual tests before handing off to AI

GAIN: A vocabulary and framework for API testing that clarifies what to build first

GAIN: A practical mapping from CRUD operations to HTTP methods for manual test design

GAIN: A decision guide for using Postman versus cURL during the initial manual phase

STEP: Cannot tell which parts of an API need testing or why || Define ‘API testing’ and establish the importance of such testing || Knows what to test and why it matters, giving AI a clear scope to generate against

STEP: Stuck at crafting the first manual test because HTTP verbs confuse the reader || Label the CRUD operations with their corresponding HTTP methods || Maps create, read, update, delete to correct verbs, so first tests are written accurately

STEP: Uncertain which tool to use for sending the first few requests || Compute the RESTful API testing with the Postman and cURL tool || Chooses Postman or cURL for the manual setup phase, preparing a pattern for later generation

NOTFOR: Experienced testers who already know how to design and run manual API test cases

NOTFOR: Teams that have fully automated their testing pipelines and no longer need baseline patterns

FACT: Average duration: 1.5-3 Hours (Contains Video)

FACT: CPD-accredited certificate available

FACT: Intermediate level

COVERS: Define ‘API testing’ and establish the importance of such testing

COVERS: Explain the client-server architecture and the meaning of web service

COVERS: Categorise the web services into REST or SOAP

COVERS: Outline the process of sending requests in JSON format

COVERS: Compute the RESTful API testing with the Postman and cURL tool

COVERS: Label the CRUD operations with their corresponding HTTP methods

COVERS: Distinguish between Postman and cURL

COVERS: Recall how to write API test cases

COVERS: Programming

COVERS: API

COVERS: Technology

MODULE: Test Rest API Manually || API Roadmap and Introduction | Manual Web API Testing and Cases

What you are probably seeing

You review the output of your automated coding agent and notice the generated tests are syntactically correct but fail to capture the edge cases that break your production code.

Your test suite grows quickly, but you spend more time deleting irrelevant tests and debugging false positives than you do verifying that the actual business logic is covered.

Why it happens

Large language models rely on statistical patterns rather than semantic understanding of your specific system, so they default to generic test structures that rarely align with complex local dependencies.

Without a manually established baseline, the model has no reference for your naming conventions, mocking strategies, or error handling priorities, leading to inconsistent and often redundant test cases.

What changed recently

2026-10-07 · best ai for testing yourself
PythonAgenticAI is now building patterns for that gap:
• Skipping manual patterns is often acceptable for side projects.
• Learning is emphasized through building and peer discussion.
• The methodology is taught in a 6-week hands-on cohort.
Source: autocomplete

Step by step

1. Manually write two or three representative test cases for a core feature to establish the exact structure, naming style, and assertion logic you prefer.

2. Annotate these manual examples with comments that explain why specific mocks or fixtures are used, giving the AI clear context for the next batch.

3. Feed these annotated examples into your AI coding agent as a few-shot prompt before asking it to generate the remaining tests for that module.

4. Review the AI-generated tests for logical consistency and remove any that test trivial code or duplicate existing coverage.

5. Run the full suite and ensure all new tests pass against the current codebase before committing them to your repository.

Ways to go further

Start here: Rest API Manual Testing For Beginners For when you are new to testing and need the basics. Open the free course on Alison ↗

Affiliate links: CHEAPUNO may earn a commission if you buy through them, at no extra cost to you.

Other problems readers raise

How to use Claude skill creator for A/B testing workflows

What price should I set for my online course?

Can free courses replace expensive tutoring and textbooks?

How to use AI to amplify technical judgment

How to learn complex systems when AI feels faster