Straight answers to the questions people ask about AI
200 curated Q&As — each one written to answer a single question clearly, then link you to the assistant, prompt or tool that goes deeper.
All questions
200 questions
An assistant answers when you ask. An agent decides what to do next and takes actions on its own — sending emails, updating records, running tools.
A token is a chunk of text — usually a word or part of a word. Models charge and measure usage in tokens, not characters.
No, ChatGPT has a free tier that covers most everyday tasks. The paid Plus plan buys you the latest model, more usage, and image and voice features.
The best early wins are content (newsletters, blog posts, product copy), customer support drafting, SOPs, and turning meeting notes into action lists.
Show, don't tell. Paste 3–5 samples of your best writing into the prompt and ask the model to match the style, not describe it.
Not reliably. Detectors have high false-positive rates on natural human writing and are easily fooled by light editing.
Fine-tuning is retraining a model on your own examples so it picks up your style or specialist knowledge. It is powerful but usually not what you actually need.
For general drafting, Claude produces the most natural long-form prose today. ChatGPT is faster and better for structured formats. For a role-tuned writer, use a SynaBot writing assistant.
A prompt is the instruction you give an AI model. Good prompts are specific about the role, the task, the format, and the constraints.
RAG is a technique where the model looks up relevant documents before answering, so its answers are grounded in your data instead of guesses.