OpenAI API basics and safe patterns

Day 10 of 7 · Build an AI App in a Week

One-liner: Make a reliable OpenAI API call with basic safeguards.
Time: 20 to 30 min
Deliverable: Working OpenAI Call with Error Handling

Learning goal

You will be able to: Implement a basic OpenAI API request with safe defaults and error handling.

Success criteria (observable)

The request returns a response for a sample input.
Errors are handled with a clear fallback message.
The API key is stored in an environment variable.

Output you will produce

Deliverable: Working OpenAI Call with Error Handling
Format: Code snippet plus test log
Where saved: Repo and course folder notes

Who

Primary persona: Digital nomad building an AI app Secondary persona(s): Early users relying on stable output Stakeholders (optional): Collaborators

What

What it is

A small, reliable server side call to the OpenAI API with safe defaults. It returns a useful response and fails in a predictable way when something breaks.

What it is not

It is not a full agent system or advanced orchestration. It is the first stable building block you can trust.

2-minute theory

Safe defaults reduce surprise behavior and cost spikes.
Clear error handling protects user trust on bad days.
Environment variables keep secrets out of the codebase.

Key terms

API key: The secret used to authenticate requests.
Fallback: A safe response when the API fails.

Where

Applies in

Server routes
Backend services

Does not apply in

Client side secret storage

Touchpoints

API route
Error logs
Env configuration

When

Use it when

You are adding AI features
You need a stable first response

Frequency

Once per product, revisit as features grow

Late signals

Frequent errors without clear messages
Secrets committed to the repo

Why it matters

Practical benefits

Faster debugging when something fails
Safer deployments without leaked keys
Better user experience when the API is slow

Risks of ignoring

Leaked keys and unexpected costs
Unclear failures that erode trust

Expectations

Improves: stability and trust
Does not guarantee: perfect outputs

How

Step-by-step method

Store the API key in an env variable.
Send a request with a short prompt.
Add a timeout and catch errors.
Return a fallback response on failure.
Log a request id for tracing.

Do and don't

Keep prompts short and testable
Log errors with request context

Don't

Put API keys in the client
Hide errors from logs

Common mistakes and fixes

Mistake: Key in code. Fix: Move to env and rotate the key.
Mistake: No fallback. Fix: Add a safe default message.

Done when

Request succeeds with a sample input.
Errors return a clear message.
API key is stored in env.

Guided exercise (10 to 15 min)

Inputs

OpenAI API key in env
A sample input prompt

Steps

Write a simple request function.
Add error handling and a fallback message.
Run a test input and record the result.

Output format

Field	Value
Input prompt
Response sample
Error handling
Fallback message

Pro tip: Log a request id so you can trace failures quickly.

Independent exercise (5 to 10 min)

Task

Change the prompt to a new input and verify output still works.

Output

Test log with the new prompt and response.

Self-check (yes/no)

Is the API key stored in env variables?
Do errors return a user safe message?
Is a fallback response defined?
Is a test run recorded?

Baseline metric (recommended)

Score: 3 of 4 checks met
Date: 2026-02-06
Tool used: Terminal

Bibliography (sources used)

OpenAI API Docs. OpenAI. 2026-02-06. Read: https://platform.openai.com/docs
OWASP API Security Top 10. OWASP. 2024-01-01. Read: https://owasp.org/www-project-api-security/