Use it for
Choosing the right AI tool for prompting, code generation, repo assistance, and local experimentation.
AI Tools Cheat Sheet
A practical setup and decision guide for the main AI tools we use in the QA and automation workflow.
Default approach
Use hosted tools for speed, local tools for privacy, and coding assistants for in-IDE flow.
Main lesson
No single tool wins every use case; select based on privacy, cost, speed, and workflow fit.
Pair it with
Use this with prompt engineering, Claude Code, Ollama, and LLM evaluation.
Hosted LLM Tools
These tools are the fastest way to start experimenting and building QA workflows.
Common tools: - ChatGPT - Claude - Gemini - DeepSeek-hosted options
When to Use ChatGPT
Good for broad prompting, structured outputs, drafting, and fast iteration.
Use for: - test plan drafting - bug summaries - requirement analysis - API test ideas
When to Use Claude
Useful for long-context reasoning, careful writing, and repo-style assistance workflows.
Use for: - long requirement review - code review - architecture notes - agent planning flows
When to Use Gemini
Good to know when comparing output quality and workflow fit across major providers.
Compare across: - accuracy - reasoning style - speed - formatting quality - tool ecosystem
Coding Assistants
Use IDE-based tools when the problem is code-shaped and context lives in the project.
Examples: - GitHub Copilot - Cursor - Claude Code - Amazon Q - Augment
Local AI Tools
Local tools matter when privacy or offline usage is more important than cloud convenience.
Common local stack: - Ollama - LM Studio - local embeddings - local vector DB
Quick Decision Rule
Choose tools by workflow, not hype.
Need fastest chat workflow? Hosted LLM Need repo-aware coding? IDE assistant Need privacy? Local LLM Need orchestration? n8n or LangFlow
Setup Checklist
Tool setup quality matters more than having many tools installed.
- account access working - API keys stored safely - editor integration tested - local models downloaded if needed - one good prompt library ready
Good Team Habit
Document which tools are approved for which kind of data and which tasks.
- public docs okay in cloud - secrets never in prompts - code review required - tool choices explained in onboarding