9 guides · free to read
Learn to use AI assistants well
Most AI advice is either hype or vague caution. These guides explain the mechanism — what actually happens when you send a message — and then show you what to do about it, with concrete before-and-after examples.
No account required, nothing gated. If you read one of these and never sign up for PhantomAI, the guide has still done its job.
A note on what you'll find here. These guides apply to AI assistants generally, including PhantomAI. Several of them are largely about where these tools fail and what you should not trust them with. We'd rather tell you that up front than have you discover it on something that matters — see what PhantomAI cannot do.
Fundamentals
How these tools actually work, in plain language. Start here if you want to understand why AI assistants behave the way they do.
How AI Assistants Actually Generate Answers
Tokens, context windows, temperature, and why an AI assistant sounds equally confident whether it is right or wrong. The mechanism, explained without maths.
Read guideLimitations of AI Assistants (With Failures You Can Reproduce)
Nine concrete failure modes of AI assistants, each with a test you can run in under a minute, why it happens, and what to do instead.
Read guideWhat Is an AI Assistant? A Plain-English Explanation
An AI assistant is not a search engine, a chatbot script, or a person. Here is what it actually is, what happens to your message when you send it, and where the boundaries are.
Read guidePractical skills
Techniques you can apply immediately, shown as before-and-after examples with the reasoning behind each one.
How to Fact-Check AI Output (A Workflow That Takes Two Minutes)
A triage system for deciding what to verify, plus the specific methods for checking citations, statistics, quotes, code, and legal or medical claims.
Read guideHow to Write Better Prompts (With Before-and-After Examples)
Nine prompting techniques shown as before-and-after pairs, each with an explanation of the mechanism that makes it work. No mystical "magic phrases."
Read guideWorkflows
Step-by-step workflows for specific jobs — studying, programming, and research — with the exact prompts to use.
How to Use AI for Programming (Without Shipping Broken Code)
Five concrete developer workflows with real prompts and code — debugging, test generation, code review, learning an unfamiliar codebase, and refactoring — plus the review checklist that matters.
Read guideHow to Use AI for Research (A Workflow With Verification Built In)
A six-stage research workflow using AI where it is reliable and avoiding it where it is not — including why you should never ask an AI assistant for citations.
Read guideHow to Use AI for Studying (Without Undermining Your Learning)
Five study workflows built on evidence-based learning techniques, with the exact prompts to use — plus the three ways students accidentally make AI harm their retention.
Read guideSafety
What to be careful about: privacy, data handling, and the limits of what you should trust.
Who writes these guides
All of them are written and maintained by Simanta Pratim Das, founder & developer of PhantomAI, based in Guwahati, India. PhantomAI is a solo project — there is no editorial team, and we don't pretend otherwise. You can see how these are written, corrected, and dated in our editorial standards, and verify the author on GitHub.
Want to practise these techniques?
PhantomAI is free to use. Bring one of the prompt patterns from these guides and try it on something you're actually working on.