You’re past the basics
If you’ve worked through the first four lessons, you know what AI does with documents, you’ve uploaded a file, you know why answers sometimes go wrong, and you have a sense of what’s safe to share. That’s enough to be genuinely useful in daily life.
Where to go from here depends on what you actually want to do. This page is a branching map. Pick the section that matches your goal, ignore the rest.
"I want to summarize or search my own documents"
Most common first use case. You have a pile of PDFs — bank statements, landlord letters, product manuals, research papers, meeting notes — and you want to ask questions across them.
- Uploading files to ChatGPT — the deeper version of what lesson 2 covered. Explains what ChatGPT is really doing behind the scenes when you attach a file, and when it works well versus poorly.
- Uploading files to Claude — Claude handles long documents better than most. Read this if you’ve got anything over 100 pages.
- NotebookLM guide — Google’s tool aimed specifically at research and note-taking. Excellent for building a "brain" from a folder of documents.
- PDF to Markdown converter — the tool itself, when you hit a PDF that ChatGPT can’t read cleanly.
"I want to turn physical papers into something I can search"
You have a file drawer, a stack of old receipts, or a shoebox of handwritten notes. The answer isn’t "open ChatGPT." The answer is: digitize first, then use AI on the results.
- Scanning old family documents — how to get good scans from photos, phones, and flatbed scanners.
- Converting handwritten notes — the one that AI tools usually fail on. The post covers the workarounds that work.
- Handwritten-to-text converter — for the single-page jobs where you just want a clean transcription.
- Digitizing old documents — the bigger project version, aimed at people doing a whole archive.
"I want to work with books or long articles"
Ebooks (EPUB), long PDFs, and research papers are their own category. The techniques that work on a two-page bill don’t always work on a 400-page novel.
- EPUB to markdown — for converting ebooks into something an AI can read.
- Build an AI assistant from books — the bigger project: turning a personal library into something you can ask questions of. Mostly for coaches and researchers.
- Research papers to NotebookLM — a specific workflow for students and academics.
- Claude uploads guide — Claude’s long context window is the most forgiving for book-length material.
"I want to translate or simplify difficult text"
One of the genuinely strongest uses of AI. Take a stiff legal page or a foreign-language pamphlet, get back something you can actually read.
- Translation works in any AI chat — just paste or upload and say "translate to English at a 10th-grade reading level."
- For legal documents specifically, the legal documents guide covers how lawyers (and non-lawyers) approach this.
- If you’re helping someone navigate a bureaucracy — insurance, benefits, immigration — ChatGPT or Claude can usually turn a dense form letter into a plain-English "what this is asking you to do" explanation.
"I want to get better at writing prompts"
The single highest-leverage skill. A better prompt beats a better model most of the time.
- Custom instructions for ChatGPT — set up your profile so ChatGPT behaves consistently across every conversation.
- Be specific. "Summarize this contract" is weak; "pull out every deadline, every dollar amount, and every party’s obligations, as a bullet list" is strong.
- Ask for format explicitly. "Give me the answer as a markdown table" or "answer in three bullet points or fewer." AI tools will honor these requests almost every time.
"I want to understand the jargon"
You’re going to keep bumping into terms — token, context window, RAG, vector, embedding, prompt, temperature, hallucination. None are as scary as they sound.
- The glossary — plain-English definitions for the AI and file-format vocabulary that shows up in articles and documentation.
- Markdown basics — if you’re tired of seeing asterisks and pound signs in AI output and wondering what they mean.
- File formats explained — PDF vs DOCX vs TXT vs markdown vs HTML, with when-to-use-what guidance.
"I work in a specific field and want the version for me"
For professionals, generic beginner content runs out fast. We have a separate track of educational guides aimed at specific fields — lawyers, accountants, teachers, healthcare admin, small business owners, coaches.
- By-industry learning hub — the index page covering all the fields we have guides for.
- Lawyers: practical AI for lawyers.
- Accountants and finance: the practical guide.
- Teachers and educators: the classroom-focused version.
- Small business owners: the operational workflow guide.
"I want to just try more tools and see what sticks"
Fair. Here’s the natural exploration path most people take from this point:
- Use ChatGPT with a dozen of your own documents across different file types.
- Hit a problem (slow on long docs, bad with tables, won’t do a task).
- Try Claude, or try Gemini, and see whether it handles that specific problem better.
- When you find an AI that works for a particular type of task, stick with it for that task and keep the other around for the others.
There is no single "best" AI. There’s "best for you, today, for this kind of document." Lean into the test-it-yourself approach; you’ll outpace anyone reading comparison articles.
The thing that matters most
Six months from now, the AI tools will look different. New models, new features, new privacy defaults. The instincts you’re building right now — be specific, verify numbers, match the tool to the document — will still apply. Tools change faster than the skills. Keep using them, keep noticing what works, and you’ll stay ahead of most people who only read about it.
Good luck. If you run into something this path didn’t cover, the main learning hub has the deeper articles, and the free tools are there when you want to try something hands-on.