Popular Gen AI Tools
Gen AI · Multimodal & Applied AI
Theory
Generative AI tools fall into a few categories, and knowing the categories matters more than memorising product names, which change quickly. General-purpose chat assistants (such as ChatGPT, Claude and Gemini) handle writing, research, analysis and Q&A and often accept files and images. Coding tools (such as GitHub Copilot, Claude Code and Cursor) live in editors or terminals and autocomplete, explain or edit code. Image tools (such as Midjourney, GPT Image in ChatGPT and Stable Diffusion) create visuals from text.
Underneath, providers also offer model APIs, and serious products are usually built by calling those APIs with custom logic around them: prompts, RAG, tool calling and evaluation. Open-weight models you can run yourself are another option when privacy or cost control matters. Choose by task, data sensitivity and integration needs, then test with your own examples.
Why we use it
Picking the right category of tool saves time and money: a chat assistant for drafting, a coding tool inside your editor, an image tool for visuals, and an API when you need to build something repeatable.
When to use it
Use consumer tools for personal productivity and quick experiments. Move to model APIs or open-weight models when you need automation, integration with your data, version control and repeatable results.
How it works
- Define the task
- Is it a one-off personal task? Yes: Use a chat, coding or image assistant directly. No: Plan to build on a model API.
- Must data stay on your servers? Yes: Consider a self-hosted open-weight model. No: Use a hosted API.
- Add prompts, RAG or tools as needed
- Test with your own examples before rollout
Example
Choosing a tool for a task:
- "Draft a marketing email" → ChatGPT / Claude / Gemini
- "Refactor this Python module and add tests" → Claude Code / Cursor / Copilot
- "Generate a hero image for a landing page" → Midjourney / GPT Image (ChatGPT) / Stable Diffusion
- "Build a custom support chatbot for my SaaS product" → an API (OpenAI/Anthropic/Google) + your own RAG pipelineKey points
- Learn the categories: chat, coding, image and media, APIs, open-weight models
- Product names and features change fast; test before you commit
- Serious products usually combine a model API with custom logic
- Open-weight models suit strict privacy or self-hosting needs
- Check your organisation's data policy before pasting work data into any tool