Notebook
Short, practical notes on AI tools, models and keeping your spending sensible.
-
Automate Your Weekly Report Draft with a Script
Build a small script that turns weekly notes into a reviewable report draft, with practical checks to keep errors and spending under control.
-
Summarising Long PDFs on a Small Budget
A practical walkthrough for extracting PDF text, testing a low-cost workflow, and building a checked summary without repeated full-document requests.
-
Build an AI Telegram Bot Without Overspending
Build a personal AI Telegram bot step by step, with secure credentials, short conversations, spending limits, and practical testing.
-
A Command-Line Translator in Fifty Lines
Build a small terminal translator step by step, with clear inputs, predictable output, and practical controls for a student-sized budget.
-
Build a Study Assistant Without Overspending
Build a personal study assistant with a chat API, grounded answers, useful quizzes, and a budget you can control.
-
Call Several AI Providers Through One Endpoint
Connect once, test models from different providers, and track each request without turning a small project into a large bill.
-
Send Images to Vision Models, Step by Step
Learn how to prepare images, build a vision API request, check the response, and control spending through MyRelay.
-
Debug Request Logs Without Leaking Secrets
A practical walkthrough for logging AI requests safely, tracing failures, and spotting retries that quietly increase your spending.
-
Concurrency Limits: A Practical Walkthrough
Understand why concurrency limits exist, then build a simple queue, handle retries, and keep parallel AI requests within your budget.
-
Retry Flaky AI Requests Without Wasting Your Budget
A step-by-step walkthrough for handling network failures with bounded retries, sensible timeouts, backoff, and spending checks.
-
Count Tokens Before You Send a Request
A practical walkthrough for estimating input tokens, reserving output room, and checking request costs before spending your API budget.
-
Function Calling Through a Relay Endpoint
Build a small function calling loop through MyRelay, validate tool requests, return results, and keep token costs predictable.
-
Use the OpenAI Node.js SDK with a Custom Endpoint
Connect the OpenAI Node.js SDK to MyRelay, verify your first request, and keep retries, tokens, and testing costs under control.
-
Keep API Keys Out of Your Git History
A practical walkthrough for storing API keys safely, checking commits, and handling leaks without putting your development budget at risk.
-
Streaming AI Responses with Server-Sent Events
Build an SSE response flow step by step, handle interruptions, and track token costs without making your first streaming project expensive.
-
Handle Rate Limits Without Wasting Your Budget
A step-by-step walkthrough for pacing AI requests, retrying safely, saving progress, and keeping script costs predictable.
-
Your First API Call with cURL, Explained
Make your first AI API request with cURL, understand what each part does, and check the result without wasting your budget.
-
Hidden Image and File Costs: A Spending Walkthrough
Follow a practical walkthrough to uncover image and file input costs, trim unnecessary content, and measure spending before scaling.
-
Keep Your Side Project Under a Coffee a Day
A practical walkthrough for setting limits, choosing models, tracking usage, and keeping AI spending predictable with MyRelay.
-
When an AI Pass Beats Paying Per Token
A step-by-step way to compare an AI pass with token billing before you spend more than planned.
-
Understanding a Pass Multiplier on Your Monthly Quota
A practical walkthrough for estimating how a pass multiplier affects your monthly AI budget and planning usage calmly.
-
Track Project Spend with Separate API Keys
A practical way to separate AI spending by project, set limits, and review costs before a small budget becomes a surprise.
-
Cheap First, Expensive Second: A Model Cascade
Build a model cascade that handles routine requests cheaply and escalates difficult work while keeping quality and total spending visible.
-
How to Plan Around Weekly Pass Caps
A practical walkthrough for managing weekly passes, choosing models, and keeping AI spending predictable with MyRelay.
-
How a Small Overdraft Floor Limits Surprise Bills
A practical walkthrough of using a small negative-balance limit, request costs, and a personal budget to keep AI spending predictable.
-
The Real Cost of Long Chat Histories
A practical walkthrough for understanding, measuring, and controlling the cost of long AI chat histories.
-
How Cache Reads Are Billed and Why They Are Cheap
Follow a request from prompt to bill, learn why cached input costs less, and check whether caching is reducing your API spending.
-
Fable vs Opus: When the Premium Model Is Worth It
A practical walkthrough for testing Fable against Opus, measuring total cost, and deciding when a premium model earns its place.
-
A Simple Rubric for Comparing Model Answers
Compare model answers with a small, repeatable rubric that checks correctness, usefulness, and cost before you spend more.
-
Compare Context Windows Without Wasting Tokens
A step-by-step way to compare model context limits, test useful recall, and choose enough room without paying for unnecessary tokens.
-
Choose a Model for Translating Documentation
A practical walkthrough for testing translation quality, preserving technical details, and keeping documentation costs predictable.
-
Pick the Right Model for Your Hackathon Weekend
A practical walkthrough for testing models, setting a spending limit, and shipping a hackathon demo without burning your budget.
-
GLM and MiniMax: Two Options Worth Testing
A budget-conscious walkthrough for testing GLM-5.3 and MiniMax M3 on real tasks, comparing results, and deciding when to use them.
-
A Practical Qwen Workflow for Multilingual Writing
Build a repeatable Qwen writing workflow with clear briefs, small tests, locale checks, and request-level cost tracking.
-
Grok for Brainstorming: A Practical Walkthrough
Use Grok to explore ideas, challenge assumptions, and choose a small experiment without letting brainstorming consume your budget.
-
DeepSeek V4 Reasoning on a Small Budget
A step-by-step walkthrough for testing DeepSeek V4, controlling token use, and checking reasoning results without stretching your budget.
-
Which Model Should Summarise Your Lecture Notes?
A practical walkthrough for choosing a model, testing summary quality and keeping lecture-note summarisation costs under control.
-
Gemini Flash for Fast Drafts and Quick Answers
A step-by-step walkthrough for using Gemini Flash to draft, ask focused questions, and manage your AI spending with MyRelay.
-
Haiku for Cheap Background Tasks: A Practical Walkthrough
Build a budget-conscious background workflow with Claude Haiku 4.5, clear task boundaries, validation, and request-level cost tracking.
-
GPT-6.1 Sol vs GPT-6 Astra: Pick One for Coding
A budget-conscious walkthrough for comparing GPT-6.1 Sol and GPT-6 Astra on real coding tasks before choosing a daily model.
-
Review Your Code with an Assistant Before You Push
A step-by-step pre-push review workflow that catches useful issues, keeps you in control, and limits unnecessary AI spending.
-
Ask a Coding Assistant to Explain Unfamiliar Code
A step-by-step walkthrough for getting clear, evidence-backed code explanations without sending an entire repository or wasting tokens.
-
Claude Opus vs Sonnet for a Solo Developer
A step-by-step workflow for choosing Sonnet or Opus, checking results, and keeping everyday AI development costs under control.
-
Refactor Legacy Code with a Coding Agent, Step by Step
A practical walkthrough for safer legacy refactoring with a coding agent, focused tests, small changes, and a controlled AI budget.
-
Gemini CLI and Long Files: Practical Limits
A step-by-step walkthrough for handling long files in Gemini CLI while controlling context, checking coverage, and keeping costs predictable.
-
Write Tests for Existing Code with Claude Code
A practical walkthrough for adding useful tests with Claude Code while keeping changes small, checking behavior, and watching API costs.
-
When to Let Your Coding Agent Run, and When to Stop
A practical walkthrough for setting coding-agent limits, spotting costly loops, and deciding when another attempt is worth paying for.
-
A Git Workflow for AI Coding Agents
A practical walkthrough for keeping AI-assisted changes small, reviewable, and affordable with branches, checkpoints, tests, and clear commits.
-
Use a Coding Assistant Safely on a Shared Campus PC
A practical walkthrough for protecting your API key, limiting costs, and cleaning up after AI coding sessions on shared university machines.
-
Codex CLI Configuration, Explained Line by Line
Walk through Codex CLI settings for models, providers, credentials, permissions, and profiles without losing track of your spending.
-
Set AI Tool Environment Variables on Any OS
Set up AI coding tool environment variables on Windows, macOS and Linux, then verify access and avoid unexpected API spending.
-
Use Claude Code Responsibly for Small Assignments
A step-by-step guide to using Claude Code for small assignments while protecting your learning, checking results, and controlling costs.
-
Pair Program with AI on a Small Monthly Budget
A practical walkthrough for planning AI coding sessions, limiting context, checking costs, and saving your budget for difficult problems.
-
How a Coding Agent Spends Tokens
Read a coding agent’s transcript step by step to understand input, tool results, repeated context, and the requests behind your bill.
-
Switch Between Claude Code and Codex in One Project
A practical walkthrough for moving between Claude Code and Codex while keeping project context, code changes, and AI spending under control.
-
Use Cherry Studio to Chat With Several Models
Set up Cherry Studio with MyRelay, add models, compare replies, and keep desktop AI chats within a personal budget.
-
Point OpenCode at MyRelay Without Losing Budget Control
Connect OpenCode to MyRelay, check your first request, and keep terminal coding sessions within a personal pay-as-you-go budget.
-
Use Claude Code in VS Code Without a Team Plan
Set up Claude Code in VS Code with personal API billing, check your connection, and keep coding costs predictable with MyRelay.
-
Gemini CLI for Quick Scripts: A Solo Setup
Set up Gemini CLI for small scripting jobs with focused prompts, safe testing, and a practical pay-as-you-go budget.
-
Run Codex CLI with a Pay-As-You-Go Endpoint
Connect Codex CLI to MyRelay, verify your first request, and keep coding sessions within a budget you can comfortably manage.