DEV Community: KD Jayakody

A free AI study workflow: lecture PDF to revision notes

KD Jayakody — Mon, 02 Nov 2026 16:23:01 +0000

I teach AI and digital skills to students in Sri Lanka, and the same thing happens every exam season. Students have plenty of material (lecture PDFs, slides, recorded classes) and very little time. Most of them read the same notes again and again, which feels productive but isn't the best way to remember anything.

This is the workflow I recommend now. It's built from free tools, and it has four steps:

Capture: get all your material into one place.
Summarise: turn it into short notes you can actually read.
Active recall: test yourself instead of re-reading.
Spaced review: come back to the hard parts at the right time.

AI helps with steps 1–3. Step 4 is where you do the remembering.

Step 1: Capture everything in one notebook

I use Gemini Notebook for this. Until July it was called NotebookLM; Google renamed it on July 16, 2026, and it's the same standalone product, now also available inside the Gemini app.

Create one notebook per subject (or per unit) and add your sources. It accepts a lot of formats:

PDF, Word (docx), PowerPoint (pptx), text and Markdown files
Google Docs and Slides
web URLs
public YouTube videos that have captions (it imports the transcript)
audio files, which it transcribes on import (Sinhala and Tamil are both on the supported list)

Free users can add up to 50 sources per notebook, and each source can be up to 500,000 words or 200 MB.

Tip: one notebook per unit works better than one giant notebook for the whole year. Answers stay focused, and it's easier to see what's missing.

Step 2: Summarise, but ask specific questions

Each source gets an automatic summary in the Source Guide. That's a good start, but the real value comes from asking narrow questions in the chat, for example:

"List the 10 key definitions in the 'Unit 3 lecture' source."
"Explain the difference between X and Y using only my sources."
"What topics in this unit appear in more than one source?"

Because the answers are grounded in your own files, you can check every claim against the page it came from. Do that. AI summaries can miss things or get them wrong, and your exam is based on your syllabus, not on what the model thinks is important.

Write the final short notes yourself, even if you copy most of the structure from the chat. Rewriting is part of learning.

If you study in Sinhala or Tamil: this is a gap I kept running into with my own students, so my team and I built shortme.lk. You upload a lecture PDF and get short notes (or a practice question paper) in English, Sinhala or Tamil. The free plan covers 3 AI documents a day, up to 10 pages per PDF. Use whichever tool fits your language. The workflow is the same.

Step 3: Active recall with flashcards and quizzes

Reading your notes tells you what's on the page. Testing yourself tells you what's in your head. That's the whole idea of active recall.

In Gemini Notebook, open the Studio panel and choose Flashcards or Quizzes. Before generating, you can:

pick the difficulty (easy, medium or hard)
add a prompt, e.g. "focus on definitions and formulas from Unit 3"

While you study, you can mark each card "Got it!" or "Missed it!", ask for an explanation, and at the end practise again with only the cards you missed. Quizzes give you hints and explanations too.

Then do one more thing: download the flashcards as a CSV file. That's the bridge to step 4.

Step 4: Spaced review in Anki

Anki is a free, open-source flashcard program. When you review a card, you rate how well you remembered it, and Anki schedules the next review for around the time you're most likely to forget it. Easy cards come back rarely and hard cards come back often.

The desktop app (Windows, macOS, Linux) is free.
AnkiDroid for Android is free.
AnkiMobile for iOS is a paid app (the purchase funds Anki's development).
AnkiWeb sync between devices is free.

Anki can import any plain-text file where fields are separated by commas, semicolons or tabs, as long as it's saved as UTF-8. So the flashcard CSV from step 3 goes straight in: File → Import, choose the CSV, map the columns to Front and Back, and pick a deck.

Tip: don't import 300 cards the night before an exam. Import a unit at a time, and do your reviews every day, even if it's only 15 minutes on the bus.

The weekly routine I suggest

When	What
After each lecture	Add the PDF/recording to that unit's notebook (Step 1)
Same week	Ask 3–5 specific questions; write your own short notes (Step 2)
End of the unit	Generate flashcards and a quiz; download the CSV (Step 3)
Every day	Do your Anki reviews (Step 4)
Before the exam	Retake the quiz with only the questions you missed

A few honest warnings

Check everything against your textbook or lecturer's notes. AI tools make mistakes, especially with local syllabus content and non-English terms.
Only upload material you're allowed to use. Google's own help page says to avoid uploading documents you don't have rights to.
The tools are the easy part. The improvement comes from testing yourself and reviewing regularly, not from having the perfect app.

If you try this, I'd like to know which step you skip most often. For most of my students it's step 4.

— KD Jayakody

Getting Started with VIDO Studio: Your First Screen-Recording Tutorial, Step by Step

KD Jayakody — Mon, 02 Nov 2026 16:22:09 +0000

My team and I built VIDO Studio at KDJ Lanka, so read this with that in mind. I'll keep it practical: this is the walkthrough I'd give a teacher or creator sitting next to me on day one.

Most screen recordings fail for the same reason. The viewer can't see what you clicked. The cursor is tiny, the screen is huge, and the important part is lost in a corner. VIDO Studio is a desktop screen recorder and video editor for Windows, macOS & Linux (X11) that tries to fix that by using what you did during the recording (clicks, typing, mouse movement) to build the edit for you.

Step 1: Download and pick a plan

Go to vido.lk and download the build for your computer: Windows, macOS (Apple Silicon) or macOS (Intel). The current version is 1.6.0.

You can start on the Free plan. It records the primary display, lets you import, edit, save and export, and adds a VIDO watermark to exports. Trial unlocks the premium tools but still keeps the watermark. Pro removes the watermark and adds AI Intro, which is still a beta. You don't need to decide on day one.

Step 2: Plan a short first recording

Before you press record, pick one small task to explain, something you can show in two or three minutes. For example: "How to create a Google Form" or "How to add a formula in Excel."

A few habits help a lot:

Close notifications and any tabs with private information.
Increase your app's zoom or font size a little so text is readable on phones.
Move your mouse slowly and click deliberately. VIDO's zooms follow your clicks, so clear clicks give clean zooms.
Write down three steps you'll cover, so you don't wander.

Step 3: Set up and record

In the recording screen, choose your display, microphone and (optionally) your webcam. Webcam quality options are Smooth (640x480) and HD (720p). System audio can be recorded on supported Windows and macOS setups, depending on your permissions and source.

VIDO runs pre-record checks and gives you a countdown. During the recording you can pause and resume, which is useful when you need to open something or catch your breath. If the app crashes mid-session, crash recovery tries to bring your recording back, but save your projects as you go.

Step 4: Let Smart Auto Zoom do the first pass

This is the feature that changes how a tutorial feels. After recording, Smart Auto Zoom builds zoom regions from your recorded activity. You can choose how aggressive it is:

Clicks Only zooms in when you click.
Clicks & Typing also zooms on typing.
Dynamic & Snappy gives faster, punchier movement.

Zoom levels range from 1.2x to 3.0x. Every generated zoom appears on the timeline, so you can move, shorten, delete or adjust any of them, and add manual zooms and pan regions where the automatic pass missed something.

Step 5: Clean up on the timeline

The timeline editor is where you tidy the story:

Split and trim the mistakes and the start/end of the recording.
Speed regions let you fast-forward through slow parts, like a page loading.
Auto Cutter (Trial/Pro) finds silent pauses and proposes cuts. Review them before you export; it's a suggestion, not a final decision.
Add text overlays for steps and warnings. Templates include Title, Step, Callout, Warning, Shortcut and Plain.

Undo, redo and snapping are there, and you can bring in more clips and images if your tutorial needs them.

Step 6: Make the cursor and frame look good

Small visual choices make a big difference:

Increase the cursor size so viewers can find it on a phone screen. Cursor themes, click effects and click sounds are premium extras.
Put your recording on a background (solid colour, gradient, wallpaper or your own image) with padding and a shadow.
Style your webcam as a circle, square or rectangle, and place it where it doesn't cover the action.

Save a combination you like as a preset so every future video looks consistent.

Step 7: Add captions

Many people watch on mute, so captions matter. VIDO can generate captions automatically in several languages, including English, Sinhala and Tamil. Be aware that auto captions use Google Speech, so they need an internet connection and your audio is sent for recognition. Recording and most editing stay on your computer.

Always read through the captions and fix names and technical words. You can choose from 10 caption styles, from Classic and Minimal to Karaoke and Word Pop, and import or export SRT files if you work with subtitles elsewhere.

Step 8: Export

Choose your canvas ratio (16:9 for YouTube, 9:16 for Reels/TikTok/Shorts, 1:1 for feeds, or 4:3 and 3:4), then export. MP4 is the normal choice. GIF is handy for short loops in docs or chat, but it has no audio.

Two honest notes about export settings. A higher export resolution or frame rate doesn't add detail that your recording didn't capture, so 1080p is a sensible default for most tutorials. And if you need several ratios, you can queue them in one go; VIDO writes the files, and you upload them yourself.

A simple first-week routine

Day 1: record a two-minute tutorial and export it with default settings.
Day 2: re-edit it, adjusting the auto zooms and cutting pauses.
Day 3: add captions and a step title, and save a preset.
Day 4: export the same project as 9:16 for a short.

By the end of the week you'll have a repeatable workflow, which matters more than any single feature.

If you try it, you can download VIDO Studio from vido.lk. I'd genuinely like to hear what confused you, because that's what we'll fix next.

— KD Jayakody

5 Sites I Use So My Vibe-Coded Apps Don't Look AI-Generated

KD Jayakody — Sat, 31 Oct 2026 20:17:10 +0000

I teach AI and web development, and the most common thing I see in vibe-coded apps is that they all look the same: same gradients, same cards, same "AI default" feel. The fix isn't a better model. It's better references and better components. Here are the five I keep coming back to.

1. 21st.dev

What it is: A large library of React + Tailwind CSS components, templates and shadcn themes made by design engineers, with a CLI (npx @21st-dev/cli add) and an MCP server so your AI agent can find and install components.
Use it when: You want polished, real-world sections (heroes, AI chat UIs, pricing, sign-ins) and you're working with Cursor, Claude Code or another MCP client.
Free vs paid: A free account can copy 2 components a day. Paid plans remove the limit.

2. React Bits (reactbits.dev)

What it is: Free animated React components: text animations, backgrounds and micro-interactions. Pick JavaScript or TypeScript, with plain CSS or Tailwind, then copy the source or install it with the CLI.
Use it when: Your UI works but feels flat and you want a bit of motion that you fully own and can tweak.
Free vs paid: The main library is free. React Bits Pro (pro.reactbits.dev) is a paid one-time purchase that adds blocks, app UI and templates.

3. Magic UI (magicui.design)

What it is: 150+ free, open-source animated components built with React, TypeScript, Tailwind CSS and Motion, designed as a companion to shadcn/ui.
Use it when: You're building a landing page and already use shadcn/ui.
Free vs paid: The components are free. Magic UI Pro (pro.magicui.design) sells landing-page templates and sections for a one-time fee.

4. Uiverse (uiverse.io)

What it is: A community-made, open-source library of UI elements such as buttons, cards, loaders and inputs. Copy them as HTML/CSS, Tailwind, React or Figma.
Use it when: You need one small element that looks hand-made, even without React.
Free vs paid: The site says it's 100% free for personal and commercial use.

5. Mobbin (mobbin.com)

What it is: Not a component library. It's a UI/UX reference library of screens from real, shipped mobile and web apps.
Use it when: You need to see how real products design onboarding, paywalls or settings before you prompt your AI.
Free vs paid: Joining is free, with limited access to apps and sites. Flows, animations and search need a paid plan, and there's a student/educator discount.

Quick tip

Don't ask your AI for "a modern UI". Pick one real reference screen from Mobbin, choose one component library (mixing five libraries is how apps start looking generic), and give the AI your brand colours, font and spacing scale. Consistency is what makes an app look designed.

What's on your list? Drop it in the comments.

— KD Jayakody