How to use ShowMeWise

What ShowMeWise does

• You record your screen while you show how to do something, and you talk while you do it.
• ShowMeWise watches that recording. It listens to what you say, reads what is on the screen, and writes a step-by-step guide.
• The guide has numbered steps, pictures with the clicked buttons marked, a short summary and questions and answers.
• You can save the guide as a web page (HTML), a PDF or a Word file.
• Everything happens on your own computer: your video is never uploaded anywhere.

Before you start

• Use a laptop or a desktop computer with Chrome, Edge or Firefox. Phones and tablets do not work. The browser AI needs Chrome or Edge; in Firefox, use Ollama.
• Your computer needs at least 8 GB of memory (RAM). 16 GB is better.
• Plug in the laptop and close programs you do not need. Making a guide is hard work for a computer.
• The first time, ShowMeWise downloads the parts it needs. The speech part is about 970 MB (about 1.6 GB for Accurate). The AI that writes is 0.8 to 3.5 GB, unless you use Ollama. These downloads happen only once.
• Size limit: on this website, one video can be up to 1000 MB and 30 minutes. Is your recording longer? Cut it into shorter parts first, or use the desktop app, which has no limit.
• Your video stays on your computer. It is never uploaded. You need no account, and it is free.

Words you will see

• AI: the computer program that writes your guide.
• Model: one version of that AI. A bigger model writes better, but it is slower and needs more memory.
• WebLLM: the AI runs inside your browser. Nothing to install.
• Ollama: a free program you can install (ollama.com) that runs the AI on your computer. Usually faster and better than the browser AI.
• OCR: reading the text on the screen.
• Transcribe: turning speech into written text.
• Export: saving the guide as a file.

Step 1. Start: choose your video

• First pick the language and the speech option. Then choose the video. ShowMeWise starts at once.
• No sound in your video? That is fine. The guide is then made from the screen only.
• Key words (optional): type names or codes from the video, separated by commas, for example QX21N, Weekly Overview. The guide spells them exactly like this, and a part where one is said or shown uses it.

123456

1. 1The video types that work: MP4, MOV and WEBM.
2. 2Drop your video here, or click here to pick it from your computer. MP4, MOV and WEBM files work.
3. 3The language people speak in the video. If it is not in the list, pick Another language. The guide is always written in English.
4. 4How the speech is heard. Fast is good for clear English. Accurate hears names, codes and mixed languages better, but takes 3 to 4 times longer on most laptops.
5. 5Key words (optional): names or codes from the video, separated by commas. The guide spells them exactly like this.
6. 6Record your screen: record here instead of choosing a file. The guide starts by itself when you stop.

Step 2. Wait while ShowMeWise reads the video

• The first time, you see the speech part downloading (for example 300 of 968 MB). Next time it is already there.
• Keep this tab open and in front. Then your computer stays awake. If the computer sleeps, the work stops.
• If you try to close or reload the tab, the browser asks first, so you do not lose the work by mistake.
• Cancel processing stops it. New video (top bar) lets you start again with another file.

12

1. 1How far it is. The time left shows up once the AI has written the first part.
2. 2Each job, one after the other: taking pictures from the video, getting the sound, writing down the speech, reading the screen, starting the AI, writing the steps, checking the steps, tidying up, and writing the summary and questions.

Step 3. Choose your settings

• From top to bottom: the name of your guide, your company name, which AI writes the guide, what the guide will contain, two options for accuracy, and the start button.
• Most people only check the name and press Generate Procedure. The other settings are already set well for you. Each one is explained below.

1

1. 1This window opens by itself when the speech and the screen have been read. The X closes it. Annotate in the top bar opens it again.

Step 3a. Name of the guide and company

• Detected tools: the programs ShowMeWise saw in your video. Only shown when it found some.

12

1. 1Document title: the heading of your guide, for example "Create and share a dashboard". Leave it empty and ShowMeWise writes a title for you.
2. 2Company name (you can skip it): shown at the top right of the saved files.

Step 3b. Which AI writes the guide: Ollama

• Models with "cloud" in the name are hidden unless you turn on Use Ollama cloud models (more powerful) under the model list. You must first read a short notice and tick I understand.
• A cloud model is bigger and better at reasoning, but it runs on Ollama's servers: the words and screenshots of your video go there. Do not use it for private or work videos. It needs internet and a free Ollama account (a Sign in button opens Ollama's page when needed), and free use has limits.
• The free cloud model is Gemma 4 (gemma4:31b-cloud): Use this adds it with one click, no terminal. Most other Ollama cloud models need paid Ollama credits. ShowMeWise never picks a cloud model by itself. While cloud models are on and you picked one, the plugin's chats and your pet use it too, and say ☁ answered in the cloud.
• Some models can also look at the pictures of your screen (for example gemma3). They give the most exact steps.

12

1. 1Ollama. If it is running on your computer, ShowMeWise finds it and picks it for you. "Running · 6 models" means it found 6 models.
2. 2Select model: the best one is already picked. You can choose a smaller one to go faster.

Step 3c. Which AI writes the guide: in the browser (WebLLM)

• If your browser cannot run it, use Chrome or Edge on a computer, or install Ollama.

1234

1. 1WebLLM: the AI runs inside your browser. Nothing to install. It downloads once and is kept for next time.
2. 2Test my computer: checks your computer and picks the model that fits it.
3. 3Browser model: Small (0.9 GB, only for chat), Medium (1 GB, the usual choice), Large (about 2 GB, needs more free memory) or Vision (3.5 GB). Test my computer picks the size that fits.
4. 4How exact and how fast this model is. If the model is too big for your computer, a warning shows here.

Step 3d. Ollama says "Not detected"

• You do not have Ollama? The Download Ollama button gets it for Windows or Mac. After installing, open Ollama (Windows key or Cmd + Space, type Ollama, Enter); you can minimise it. Check if it worked tells you. Still not found? Look for the llama icon, open 127.0.0.1:11434 (it should say Ollama is running), restart the computer, or allow Ollama in your antivirus. Or just use the browser AI.

12

1. 1Not detected: Ollama is not running, or it does not allow this website yet.
2. 2Follow the lit step in the Set up Ollama box: Download Ollama for your computer, open it, then Connect (one click with the ShowMeWise plugin, no terminal).

Step 3e. What the guide will contain

1

1. 1Every guide has the same parts: contents, purpose, what you need before you start, numbered steps with results and pictures, things to notice, a summary, questions and answers, and a flow chart. Parts of the video where nothing happens are left out.

Step 3f. Accuracy options

• Consolidate is on, Self-verify is off. Switch Self-verify on only if a guide has steps that never happened in the video.

12

1. 1Self-verify each step: off by default. The AI checks every step a second time and removes what it cannot see. In our tests it also removed correct details, and it takes about a third longer.
2. 2Consolidate procedure: a last tidy-up of the step names. Steps that say the same thing are joined.

Step 3g. Generate Procedure

12

1. 1How long it may take on this computer. The real time left shows once writing starts.
2. 2Generate Procedure starts writing. It is grey when Ollama is picked but not found.

Step 4. Watch it write

• You can already read the first steps while the rest is written.
• How long? A 1 minute video takes a few minutes. With the most accurate model, a 10 minute video took about 1¾ hours on a normal 16 GB laptop in our tests. A smaller model is faster.

123

1. 1The time left, measured on your computer once the first part is written.
2. 2Which part it is writing now. Your steps appear below while it works.
3. 3Cancel processing stops it. What is already written stays.

The progress list: what each line means (Skipped, kept from the last try, times)

• Each line is one job, with what came out of it on the right and how long it took. Why does a job skip? It is switched off or not needed, see below.
• Self-Verify Steps: Skipped means it is switched off. It is off by default, because in our tests it also removed correct details and took about a third longer. Switch it on in the AI settings (Self-verify each step) if you want it. Nothing is missing because of it.
• Consolidate Procedure: Skipped means it is switched off, or the guide has only one part, so there is nothing to tidy or join.
• Transcribe Audio: Skipped means the video has no sound, or the browser could not read it. The guide is written from the screen only.
• Purpose · Summary · FAQ · Flow: skipped means the AI gave no usable summary this time. Your steps are fine. Press Re-annotate to try again.
• "kept from the last try" (with 0s): you chose the same video with the same speech settings as your last run (finished or not). The speech and the screen text were saved then, so they are not read again. That is why it takes 0 seconds.
• "Picking up where it stopped: ... parts already written": parts written in the last try are kept; only the rest is written now.
• ShowMeWise knows it is the same video when the file name, the exact size, the time the file was last saved, the speech language and the Fast or Accurate choice all match. A new recording (even with one extra sentence) is a new file, so it starts fresh. Written parts are only kept with the same AI model, settings and key words.
• "... entries annotated": how many parts the guide has. "... segments": how many pieces of speech were written down. "... frames": how many pictures were taken from the video.
• A very long time, for example 356m, or "computer slept 5h": the work waits while the computer sleeps and goes on when it wakes. The time the computer slept is not counted and is shown on its own. Keep the laptop open, plugged in and the tab in front.
• "Ollama · gemma3..." with 0s next to Load AI Model: Ollama loads the model itself, so this line is done at once. With the browser AI it shows the download and loading.

Step 5. When it is done

• Scroll down to read your guide, the summary, the questions and answers, and to save it.

123

1. 1COMPLETE: every job has a green tick, with how long it took.
2. 2Re-annotate: write the guide again, for example with another model. The video is not read again.
3. 3New video: start again with another recording. Save your guide first.

Step 5a. Read and fix your guide

• The AI can make mistakes. Read every step before you share the guide.
• Every step has Edit, Move up, Move down and Delete.
• A step marked "check this" is one the AI was less sure about. Look at it first.
• Said in the video, but not in this step: something you said (and is on screen) is missing from the step. Press + Add to put it in, or Hide.
• At the top: how many steps need a quick look. Show only these lists just those steps.
• Check these short codes: if the same code is written in different ways, fix it once with Replace all, or press Looks good.
• On the left: your video and small pictures from it. Click a picture to jump to that moment in the video.

13452

1. 1All actions of this step, in the right order. Codes and numbers are copied from the screen.
2. 2Pictures of the screen. The button or field you clicked has a number. Click a picture to make it big, click again to close it.
3. 3Edit, Move up, Move down and Delete: on every step.
4. 4Self-check: how well the guide matches the video (100 is best), and how many parts it has.
5. 5Your video and small pictures from it. Click a picture to jump to that moment.

Step 5b. Change a step

• Press Save changes (or Ctrl + Enter). Cancel (or Esc) keeps the old version. Your changes are in every saved file.

123

1. 1Step title: the name of the step.
2. 2What to do: one action per box. Remove deletes an action. + Add an action adds a new one.
3. 3What you get and Tip: both optional.

Step 5c. Summary, questions and flow chart

• Above them: the purpose of the guide and a short summary. All of this is also in the saved files.

12

1. 1Flow diagram: all parts of the process from start to end. It follows your step edits by itself. In Edit document you can rename, add or remove boxes and phases.
2. 2Questions and answers about your process. Click a question to open its answer.

Step 6. Save your guide (export)

• The files are made on your computer and go to your Downloads folder. Nothing is uploaded.
• Videos in the files: only the HTML file has moving clips (click one to see it big). Word has short GIFs in some parts. A PDF cannot play videos, so it has still pictures only.

123

1. 1HTML: a web page in one file. Pictures with marks, short clips of each step, a checklist that remembers your ticks, and questions and answers.
2. 2PDF: ready to print, with a cover page, contents and page numbers.
3. 3DOCX: the same guide as a Word file you can change.

Step 7. Ask questions about the video (Vid Chat)

• Each answer shows the marked screenshot of the steps it names: only the buttons and fields the answer talks about are marked. Click a picture to make it big.
• Vid Chat works when the guide is done, or already when the speech has been written down.
• It only answers from your video. If the answer is not in it, it tells you, and offers Want an answer anyway?
• Search the web opens the search site Perplexity in a new tab, with only the question you typed. Your video and guide are not sent.
• Add a document (top right of Vid Chat): add a manual, notes or a PDF that belongs to the video. Answers then come from the video and the document together. The document is read on your computer and is not uploaded.

1234

1. 1The answer shows which step it comes from, with its picture and the time; click the time to play that moment.
2. 2Search the web for more: opens a web search with your question only.
3. 3Not in your video? It says so, and offers Answer from the AI's knowledge or Search the web.
4. 4Type a question in your own words. The answer shows which step it comes from, with its picture and a play button.

Test my computer: which browser AI size runs best

• No Ollama? ShowMeWise can use an AI that runs inside your browser. Test my computer finds the size that runs well on your computer, so nothing gets stuck. Open it from the home page, the upload page, the AI settings or the plugin.
• It tries the sizes one by one, small first: Small (Llama 3.2 1B), Medium (Qwen2.5 1.5B), Large (Qwen2.5 3B). For each one you see how fast it is and if it answered a test question right. You decide whether to try the next, bigger one.
• Bigger gives better answers but needs more free memory. Small is quick, but its answers are often weak.
• Many laptops have built-in graphics: then the AI uses the normal memory. Close tabs and programs you do not need before the test. With the ShowMeWise plugin, the test shows your free memory and how many tabs are open.
• The sizes you do not need are deleted after the test, so they do not take space. The best one is kept and used from then on.
• Ollama on your computer is still faster and more exact than the browser AI.

Record your screen (recorder)

• No recording yet? Press Record your screen on the upload page. Choose the whole screen, one app window or a browser tab, then do the task and explain it out loud. Press Stop when you are done, or Stop sharing in the browser bar. Then watch the recording, save it if you want, and press Make a guide.
• The guide starts by itself after you stop. The recording stays on your computer.
• Allow the microphone when the browser asks, so your voice is recorded. If it is blocked, the video has no voice and the guide is written from the screen only.
• On this website a recording stops by itself at the size limit. If the tab closes while recording, the next visit offers the recording again (Use it or Delete it).

The ShowMeWise browser plugin (extension) and Vidi in the plugin

• The ShowMeWise plugin for Chrome and Edge puts ShowMeWise in a small window under its icon in the toolbar. Open on the side keeps it next to your page.
• Online or offline: the top bar of every page shows 🟢 Online or ⚪ Offline, next to the ShowMeWise name. After the first visit the website opens without internet too. If you chose a cloud model and there is no internet, or its free amount is used up, the AI on your computer answers instead and a note says so (the pet tells you too). A model on your computer never switches to the cloud by itself. Tell your pet “change the AI” (or “AI badlo”) to pick one yourself: ☁ Online or 💻 Offline.
• Get the plugin free on the Chrome Web Store: search ShowMeWise there, or use the Get the browser plugin link in the top menu of the home page. Press Add to Chrome (in Edge: Allow extensions from other stores, then Add to Chrome).
• Record your screen from the plugin: the whole screen, any app window or a tab. When you stop, the guide is written by itself with the AI you used last time, and you get a notification when it is ready.
• Before you record, open Key words, language and speech in the plugin (optional): type the names and codes said in the video, pick the spoken language, and Fast or Accurate speech. Same as on the upload page; the plugin remembers them.
• The plugin shows the steps and the time left while ShowMeWise works. It keeps the computer awake while it records or writes.
• Ask Vidi in the plugin: about your video, about the page you are on, about a document (PDF, Word, web page or text), or Help for questions about ShowMeWise itself.
• This page: if the answer is not on the page, Vidi reads the links on it that match your question (same site, up to 3) and says which pages it read. Reading pages needs your OK once.
• The plugin uses the ShowMeWise website in a tab of its own, so the models are downloaded only once and an unfinished guide picks up where it stopped. If that tab is closed, press Continue in the plugin.
• Nothing is uploaded: the plugin and the website run on your computer. Only with a cloud AI (Ollama's cloud) your question and the page text it needs go to Ollama.

Your pet in the plugin (BETA): a little helper on every page

• Your pet is a small animal at the edge of your web pages: a dragon, wolf, cat, fox or owl, or your own picture or 3D pet. Its name is Vidi, and you can rename it.
• Turn it on in the plugin under Your pet. It needs the same OK as reading pages. Turn it off there, or press Hide on this site in its chat.
• Click the pet to chat. Ask about the page you are on, about ShowMeWise, or anything else. Say search the web for ... and it answers from Wikipedia and DuckDuckGo with links you can open.
• Timers, alarms and reminders: say timer 10 minutes, alarm at 7:30 or remind me at 17:00 to call Mum. It rings with a sound and a notification, but only while the browser is open. Say show my timers or cancel the alarm.
• It remembers you: say my name is ... or I love .... Say your name is ... to rename it. Forget what my pet knows about me in the plugin clears it.
• It moves away from buttons and text by itself, hides half of itself when nobody needs it, and sleeps after two quiet minutes. Drag it anywhere: that becomes its fixed spot (Let my pet find its own spot, in the plugin, undoes it).
• Your own pet: choose My picture and pick a photo, or My 3D pet and pick a .glb file. The plugin shows a text you can give a 3D maker to create one.
• Fill a form from your CV: give it the file with 📎, say fill this page, check the list (type what is missing), press Fill the page. It never presses Submit and never fills passwords, PINs or card numbers.
• Show me on the page: under an answer about a page, it marks the buttons beside them for 10 seconds and puts a screenshot in the chat.
• First steps: on the Pet training page, Meet your pet asks your name, which pet and colour, its name, the language and which AI should answer (each with how often it was right in our tests). If you pick Ollama and do not have it, it shows how to install it, step by step, and gets the model for you. When the browser opens, your pet says hello for the time of day.
• Guide me through a tool: give it the tool's manual with 📎, then ask in your own words, like how do I make a new board?. It says where you are and the first step, marked on the page. Do it yourself and press Done, next step. When it cannot tell what you want, it asks. It never clicks for you.
• Focus mode: say hide the ads, and show everything to undo. It only hides them on that page.
• Choose the AI: say change the AI. A cloud model is used only after you said your work is not confidential.
• Learn every skill on the Pet training page (#/pet-training), with made-up practice material.
• It is a test feature: answers can be wrong. Do not rely on its reminders for anything important.

Step 8. Ask questions about a document (Doc Chat)

• Doc Chat needs no video. Open it with Doc Chat in the top bar.
• If Ollama runs on your computer, it is picked for you with a quick model for chatting. You can switch to the browser AI.
• Ask anything about the document: what it is about, how to do a step, a name or a number in it.

1

1. 1Choose a document (PDF, Word, web page, text, Markdown, CSV or JSON) and ask questions about it.

Help from Vidi

• Vidi (Latin for "I saw") is the helper of ShowMeWise. Ask it anything about using ShowMeWise, in your own words.
• It knows which page you are on, so its answers fit what you see.
• Not sure which AI to use? Ask "Which model should I pick?" or "test my system". Vidi checks this computer and lists the AIs as buttons: ✅ Suggested (runs well here) first, with the most accurate on top, ⚠️ Not suggested after. Each shows its size, how big or quick it is and how often it was right in our own tests (yours can differ). Use picks it for Vidi and the app; Download gets it through Ollama with a progress bar. Without Ollama, Vidi shows how to install it, step by step.
• It watches for error messages. When one appears, a small bubble asks: "I can see an error on your screen. Do you need help with it?" Press Yes and Vidi explains the error and what to do. Press No to close the bubble. You can then keep asking.
• Do not understand a message? Copy it into the chat, or take a screenshot (Windows key + Shift + S) and paste it with Ctrl + V. Vidi reads the text in the picture on your computer.
• How Vidi answers: without any AI it shows the right part of this page at once. For written answers it uses Ollama if you have it, or a small browser AI (about 1 GB, downloaded once). Change how Vidi answers is at the bottom of the chat.
• Vidi only talks about ShowMeWise. For questions about what is in your own video, use Vid Chat. For a document, use Doc Chat.
• The first time, Vidi asks what to call you (you can say No thanks). It remembers your name and what you asked about last, only in your browser on this computer, to greet you next time. Forget me (at the bottom of the chat) clears it.

123

1. 1Vidi's chat window. It opens with the Ask Vidi button at the bottom right of every page.
2. 2Your questions on the right, Vidi's answers on the left, often with the matching picture from this page and a link to it.
3. 3Type your question, or paste an error message or a screenshot here.

Error messages and what to do

• "This video is ... MB — the online version accepts up to ... MB" or "This recording is ... min": your video is too big or too long for the website. Cut it into shorter parts (any free video editor can do that), then upload the parts one by one.
• "That file is not a video" or "This file could not be read as a video": you picked something that is not a video, or the file is broken. Pick an MP4, MOV or WEBM recording. Can it play in your video player? If not, record it again.
• "No sound in this video" (next to Extract Audio): the video has no sound, or the browser cannot read its sound. ShowMeWise makes the guide from the screen only. Did you talk in the video? Then save it again as MP4 (most screen recorders and editors can do that) and upload it again.
• "The graphics card could not load the speech model, using the processor instead (slower)": nothing is wrong. Your graphics card could not take the Accurate speech model, so the normal processor does it. It only takes longer.
• "Processing cancelled. Whatever finished is kept below": you pressed Cancel. The steps that were already written stay. Press Annotate to write the rest, or New video to start over.
• "Annotation failed ... Check the selected model is installed (Ollama) or try WebLLM": the AI could not write. With Ollama: is Ollama still open? Is the model installed (ollama run gemma3)? Or pick WebLLM instead.
• "Ollama not reachable": the page cannot talk to Ollama. Open Ollama, and do Step 3d once.
• "Ollama refused the request": Ollama got the request but could not do it, for example because the model is missing or the computer is out of memory. Check the model name, close other programs, or pick a smaller model.
• "WebLLM failed to load": the browser AI could not start. Use Chrome or Edge on a computer, close other tabs, and pick a smaller browser model (Fast). Updating your graphics driver can help.
• "Not detected" next to Ollama: see Step 3d.
• "HTML export failed" (or PDF, DOCX): saving did not work. Try again. If it fails again, try another format, or reload the page after saving what you can.
• "Leave site? Changes you made may not be saved": the browser asks this because ShowMeWise is still working. Press Cancel to stay. If you leave, the work is lost.
• Chrome asks to let the site "access other apps and services on this device": this is Chrome asking if the page may talk to Ollama on your computer. Press Allow. Your data still stays on your computer.
• "This copy only has the English speech model" (desktop app): run the setup again with the option for all languages.
• Doc Chat says "This file type does not work here" or "Could not read any text from that file": use a .pdf, .docx, .html, .txt, .md, .csv or .json file. A scanned document (a picture of text) has no text to read.
• Vid Chat says "I couldn't find this in it": the answer is not in your video or guide. Ask in other words, press Answer from the AI's knowledge, or Search the web.
• "... could not be shown. Your guide is safe and you can still export it": one part of the page had a problem, the rest still works. Press Try again next to the message. Your guide is not lost.

If something goes wrong

• "This file could not be read as a video": the file is not a video, or it is broken. Choose an MP4, MOV or WEBM screen recording.
• "Not detected" next to Ollama: see Step 3d.
• It is very slow: pick a smaller model, keep Self-verify off, close other programs and plug in the laptop. Cut a long video into shorter parts.
• The computer went to sleep, the tab was closed or the browser crashed: choose the same video again with the same settings. ShowMeWise picks up where it stopped: speech and screen are not read again, and parts already written are kept.
• A step is wrong or missing: change it with Edit. Next time, try a bigger model or Accurate speech.
• The video is too big: the website takes up to the size shown on the upload page. Cut the video into parts.

Limits, limitations and accuracy: how accurate is it, mistakes, why it works this way (free and private)

• Why it works this way: ShowMeWise is free and runs completely on your own computer. No cloud, no account, no cost per video. The AI is an open model on your machine, so the speed and the result depend on your computer and the model you pick.
• How accurate is it? We test every release against manuals written by hand for real screen recordings and read each guide step by step. With gemma3 12B on a normal 16 GB laptop (Accurate speech, key words typed in), 60 to 67.5 % of all actions came out fully right in the first draft.
• A bigger model gives a better guide: a small 4B model got about 30 % in our tests. Pick the biggest model your computer can run. The browser AI is quick to start but smaller than Ollama. In the browser, on our 1-minute demo video: Medium found 11 of 15 steps (6 fully right) in about 10 minutes; Small found only 2, so it is too weak for guides.
• Big paid cloud AIs are usually faster and stronger, but your video would have to leave your computer. ShowMeWise keeps it with you.
• Your computer sets the speed: with the biggest model a 10 minute video took about 1¾ hours on a normal 16 GB laptop. A stronger computer or a smaller model is faster. Keep the computer awake while it works.
• So a guide is a strong first draft that you finish. Read it and fix it with Edit before you share it. Start with the steps that need a quick look.
• Often missed: a click where nothing visible changes, and a button with only an icon and no text. Small clicks that were seen go to the checklist, not into the guide.
• Help it with key words: type the names and codes of your app into the key words box. In our tests this helped the most.
• Each run is a little different: an AI never writes the same text twice. In our tests the same video scored up to 10 points higher or lower.
• The guide is written in English for now, not in Hindi, German or another language, even when the video is in that language.
• Made for computers: Chrome, Edge or Firefox on a laptop or desktop. The plugin and the browser AI need Chrome or Edge; in Firefox, use Ollama. Phones and tablets are not strong enough to run the AI on the device.
• Saved files: the HTML guide has moving clips. Word has short GIFs in some parts, PDF has still pictures only, no videos. Doc Chat and Vid Chat read PDF files, but not scanned pages (pictures of text).
• Size limit: on this website, one video can be up to 1000 MB and 30 minutes. Is your recording longer? Cut it into shorter parts first, or use the desktop app, which has no limit.

Privacy and your data (short version)

• Your video, its transcript and the guide never leave your computer. Nobody can see them, not even us.
• No account, no sign-up, no analytics, no ads, no tracking cookies. We do not sell or share any data, because we do not collect any.
• The website comes from Netlify, our host. Like every host, Netlify may keep normal server logs (for example your IP address and the time).
• The AI models are downloaded once, mainly from Hugging Face, and kept in your browser. Those servers only see a normal download, never your video.
• Ollama runs on your own computer; the page talks to it there. Chrome may ask to allow local network access: press Allow.
• Search the web only runs when you click it. It opens Perplexity in a new tab with just your question.
• Vidi reads error messages and pasted screenshots on your computer only.
• Stored in your browser: downloaded models, checklist ticks of a saved HTML guide, how Vidi answers, and the transcript and screen text of the last video, so the same video starts faster (another video replaces them; parts of an unfinished guide are kept until it is done). Clear the site data in your browser settings to remove it all.
• Under data protection law (GDPR) you have rights over your data. ShowMeWise holds no personal data about you. For hosting logs, contact Netlify.
• The full text is on the Privacy, terms and credits page (Privacy in the top bar).

Terms of use and rules (short version)

• ShowMeWise is free. It is a personal portfolio project, provided as it is, without any warranty. It may change, pause or stop at any time.
• The AI can be wrong or incomplete. Check every guide before you share it or rely on it. You are responsible for the guides you make and share.
• Only use recordings you are allowed to use. Remove personal or confidential information before you share a guide.
• Do not use ShowMeWise for anything illegal or to harm others.
• Making a guide works your computer hard for a long time. Plug in the laptop, close programs you do not need, and save your other work first.
• Size limit: on this website, one video can be up to 1000 MB and 30 minutes. Is your recording longer? Cut it into shorter parts first, or use the desktop app, which has no limit.
• As far as the law allows, there is no liability for any loss or damage from using ShowMeWise.
• The AI models have their own licenses (for example Gemma by Google, Qwen by Alibaba Cloud, Phi by Microsoft, Llama by Meta). Using a model means you accept its license.
• ShowMeWise is open source. Questions, ideas or problems: open an issue on GitHub.
• The terms can change; the date on the Privacy, terms and credits page shows the latest version.

Questions people ask

• Is it free? Yes. No account, no payment.
• Is my video uploaded? No. Your video stays on your computer. Only if you pick a cloud AI (Ollama's cloud), the text and screenshots it needs to write the guide go to Ollama.
• Can I use a long video, like 30, 45 or 60 minutes, or 2 hours? On this website one video can be up to 1000 MB and 30 minutes, so cut a longer video into parts. Long videos take long. With the most accurate model, a 10 minute video took about 1¾ hours on a normal 16 GB laptop, so a 45 minute video can take many hours. Cut it into shorter parts, or pick a smaller model to go faster.
• Can I close the laptop lid while it works? No. The computer goes to sleep and the work stops. Keep it open and plugged in.
• Can I use another tab meanwhile? Yes, but keep the ShowMeWise tab open. It works best when its tab is in front.
• Which languages? The video can be in English, Hindi or Hinglish, German, Spanish, French or another language. The guide is in English.
• Can I change the guide after it is written? Yes, with Edit on every step, before you save it.
• Why does it say Skipped? A job that is switched off (Self-verify is off by default) or not needed. See The progress list.
• Why 0s and "kept from the last try"? It is the same video as your last run, so the speech and screen text were reused.