1. Overview

KnowledgeNinja is an AI-powered research and learning assistant designed to help students, educators, and professionals analyze content, ask questions, explore topics, and interact with expert personas. It combines GPT-powered chat, document analysis tools, expert role selection, voice input and audio playback, multi-language support, and custom instructions and teaching aids. Its goal is to deliver deeper reasoning, richer explanations, and more interactive learning.

2. Getting Started

2.1 Enter Your Name

Enter your name at the top of the interface. KnowledgeNinja uses this to personalize responses.

2.2 Select Your Language

Choose your preferred language from the dropdown. This affects both text and voice interactions.

2.3 Choose an AI Voice (Optional)

Select a voice such as Alloy for audio playback. If you do not select a voice, KnowledgeNinja uses Alloy by default. Audio playback is optional — the app always works fully in text.

3. Expert Roles

KnowledgeNinja includes a dropdown of 60+ expert personas, including historical figures, scientists, political leaders, educators, and subject‑matter specialists.

3.1 Use the Default Role

If you don’t select anything, KnowledgeNinja behaves as a helpful general assistant.

3.2 Choose a Predefined Role

Select a persona from the dropdown to shape the AI’s tone, expertise, and teaching style. For best results, after selecting a role, first ask: “Who are you and what do you do.” This ensures the AI fully adopts the selected persona before you begin.

3.3 Enter a Custom Role

Use the Custom Instructions field to define your own role or behavior. When switching to a new role, refresh the page first. Leftover chat from previous roles can confuse the AI.

4. Teaching Aid (Document Analysis)

4.1 Paste Text

Paste any text directly into the box.

4.2 Drag & Drop Documents

Upload files such as PDFs, text files, articles, or notes. KnowledgeNinja will analyze them and incorporate the content into its responses.

4.3 URLs

URLs are not supported as a Teaching Aid input in this app. To use webpage content, copy/paste the text instead.

4.4 How It Works

All of the following are combined into the final prompt sent to the AI: your main question or message, the selected expert role (if any), custom instructions, and Teaching Aid content (pasted text and/or uploaded documents). This merged context is what drives the quality and relevance of the response.

5. Chatting With KnowledgeNinja

5.1 Ask Questions

Use the main chat input to ask anything: homework help, research questions, document analysis, concept explanations, writing assistance, and more.

5.2 Voice Input (Optional)

Use Hold to Record to speak your question instead of typing.

5.3 Audio Playback (Optional, With Default Voice)

Every response always appears as text. There is an Audio Playback checkbox that you can turn on or off at any time. When the checkbox is ON, KnowledgeNinja speaks the response aloud. When the checkbox is OFF, responses remain text‑only. If you have not selected a voice, KnowledgeNinja uses the default voice “Alloy.”

6. Memory Controls

6.1 Clear Memory

Use the Clear Memory button to reset the session. This wipes chat history, Teaching Aid context, role selections, and custom instructions. Useful when switching topics or starting fresh.

7. Features Summary

  • GPT‑4.1 AI Chat — Balanced reasoning and responsive answers
  • Document Analysis — Paste or upload documents for context
  • Expert Roles — 60+ selectable personas
  • Custom Instructions — Define your own AI behavior
  • Voice Recognition — Ask questions using your microphone
  • Audio Playback — Optional; uses Alloy by default
  • Multi‑Language Support — Interface and responses adapt to your language
  • Research Assistant — Designed for students and educators
  • Code Analysis — Can analyze and explain code snippets

8. Tips for Best Results

  • Provide clear questions for deeper reasoning.
  • Use the Teaching Aid for long or complex documents.
  • Switch roles when you need a different perspective.
  • Use Custom Instructions to fine‑tune behavior.
  • Clear memory when changing topics to avoid mixed context.

9. Troubleshooting

  • The AI seems confused: Clear memory and re‑enter your context.
  • Document analysis isn’t working: Ensure the file is readable (PDF, TXT, etc.).
  • Voice input isn’t responding: Check browser microphone permissions.
  • Audio playback is silent: Ensure the Audio Playback checkbox is ON, ensure a voice is selected (or rely on default Alloy), and check system/browser volume.

Globalingo User Manual

Before You Begin

Globalingo only works when two or more browsers are connected to the same room. If you open it alone, you will only see your own messages — this is normal and expected. You must also be logged into the Unlocking AI Launch Platform to access the app.

1. What Globalingo Does

Globalingo enables real-time multilingual communication between browsers. Each participant speaks in their own language and hears/reads the other person in their chosen language.

2. How It Works

  1. Your browser records your speech.
  2. Audio is sent to Whisper, which returns English text regardless of the spoken language.
  3. The English text is broadcast to all browsers in the same room via WebSocket.
  4. Each receiving browser sends that English text to a GPT‑4.1 translation API call.
  5. The browser receives translated text in the user’s chosen language.
  6. The browser generates translated audio using tts‑1 and displays the translated text in chat.

3. Why Two Browsers Are Required

Globalingo is a browser‑to‑browser interpreter. If you are alone in a room, you will only see your own messages. To test properly, open Globalingo on two devices or have another person join the same room number.

4. Joining a Room

  1. Log into the Unlocking AI Launch Platform.
  2. Open Globalingo.
  3. Select your preferred language.
  4. Use the built‑in random room number generator or type any room number manually.
  5. Click “Join Room.”

Anyone using the same room number will share the conversation space. Rooms are ephemeral and have no stored history.

5. Speaking and Translating

Press and hold the Record button, speak normally, then release. Your browser handles transcription, broadcasting, translation, TTS, and display automatically.

Alternatively, you may type messages — Globalingo functions as a texting app as well; participants can send and receive text instead of audio.

6. Privacy and Data Flow

  • You must be logged into the Unlocking AI Launch Platform.
  • Globalingo does not store chat history.
  • The server relays English text only.
  • Translation and TTS occur client‑side in each browser.

7. Supported Languages

Any language supported by Whisper, GPT‑4.1, and tts‑1 is available. Each user chooses their own target language.

8. Troubleshooting

  • Only seeing your own messages: You are alone in the room.
  • No audio: Check microphone and autoplay permissions.
  • Other person can’t hear you: Verify matching room numbers.
  • Slow translations: Network or API latency.

9. Best Practices

  • Use headphones to avoid echo.
  • Speak clearly.
  • Use shorter sentences for faster translation.
  • Confirm both users selected correct languages.

10. Summary

Globalingo is a real‑time, browser‑to‑browser interpreter running inside the Unlocking AI Launch Platform. Whisper transcribes speech to English, WebSockets broadcast it, GPT‑4.1 translates it, and tts‑1 speaks it. One browser alone shows only your own messages; two or more browsers in the same room enable full multilingual conversation.

PDF OCR Processor User Manual

Before You Begin

The PDF OCR Processor requires a valid login to the Unlocking AI Launch Platform, a PDF containing scanned or image-based content, and a file size under 50MB. This tool uses OCRmyPDF and Tesseract OCR to add a searchable text layer to your PDF.

1. What the PDF OCR Processor Does

This tool converts image-based PDFs into searchable PDFs. The output looks identical to the original but gains a hidden text layer that allows text selection, copying, and searching.

2. How It Works

  1. You upload a PDF.
  2. The server processes it using OCRmyPDF and Tesseract OCR.
  3. A new PDF is generated with a searchable text layer.
  4. You download the processed PDF.

3. Supported Files

  • PDF only
  • Maximum size: 50MB
  • Multi-page PDFs supported
  • Scanned, photographed, and image-based PDFs all work

4. Using the PDF OCR Processor

  1. Log into the Unlocking AI Launch Platform.
  2. Open the PDF OCR Processor.
  3. Drag and drop your PDF into the upload box.
  4. Wait for processing to complete.
  5. Download your searchable PDF.

Note: Long PDFs may take additional processing time, especially documents with many scanned pages.

5. Best Practices for Clean OCR Results

  • Use high-resolution scans (300 DPI recommended).
  • Avoid skewed or rotated pages.
  • Ensure text is sharp and readable.
  • Prefer grayscale or black-and-white scans.
  • Minimize handwriting or annotations when possible.

6. Troubleshooting

  • PDF didn’t change: It may already contain a text layer.
  • OCR missed text: Low-quality scans or handwriting reduce accuracy.
  • File won’t upload: Ensure it’s under 50MB and in PDF format.
  • Output looks distorted: Rescan at higher resolution.

7. Privacy and Data Handling

  • Files are processed inside the Unlocking AI platform.
  • No OCR results are stored.
  • No text is logged.
  • PDFs are deleted after processing.
  • Only you can download your output.

8. Summary

The PDF OCR Processor converts scanned or image-based PDFs into searchable, machine-readable documents using OCRmyPDF and Tesseract OCR. It is a foundational utility for preparing documents for analysis in other Unlocking AI apps.

InfoNinja User Manual

Before You Begin

InfoNinja requires a valid login to the Unlocking AI Launch Platform, a stable internet connection, and URLs or documents you want analyzed. It extracts content, generates citations, organizes findings, and produces clean PDF summaries through a conversational interface.

1. What InfoNinja Does

InfoNinja is a research workflow assistant. It can analyze URLs, process uploaded documents, extract text from image-based PDFs for analysis, generate citations, organize research, export PDFs, and enhance answers using Smart Search Augmentation.

2. How It Works

Step 1 — You provide content: Paste a URL, upload a document, drop multiple files, or type directly into the chat.

Step 2 — InfoNinja extracts and cleans text: URLs are scraped and cleaned; documents are parsed with page numbers preserved when possible.

OCR Behavior for Image-Based PDFs

If you upload a scanned or image-only PDF, InfoNinja performs OCR internally to extract text for analysis. It does not modify the PDF, add a searchable text layer, or generate a new PDF.

3. Smart Search Augmentation

Every time you ask a question, InfoNinja:

  1. Generates an optimized Google-style search query based on your question.
  2. Performs a custom web search asynchronously.
  3. Pulls in relevant results.
  4. Feeds those results into the final reasoning context.
  5. Responds with text that may include hyperlinks to its sources.

This produces answers that are more accurate, current, and grounded.

4. Analysis Pipeline

InfoNinja blends your uploaded documents, URL content, extracted text (including OCR results), Smart Search results, and conversation history into a single coherent answer.

5. Supported Inputs

  • URLs (articles, blogs, documentation, research pages)
  • PDF files (text-based or image-based)
  • Text files
  • Markdown
  • Raw text pasted into chat

Unsupported:

  • Paywalled content requiring login
  • Sites that block scraping
  • PDFs that require a searchable text layer output (InfoNinja extracts text but does not rebuild PDFs)

6. Using InfoNinja

Analyze a URL: Paste a URL, press Enter, and ask questions.

Upload a document: Drag and drop a file, then request summaries, citations, or explanations.

Chat directly: Ask for summaries, explanations, citations, or PDF exports.

7. Citation Generation

InfoNinja can generate MLA, APA, and Chicago citations. Page-numbered citations are included when page data exists. URL citations include title, author (if available), and access date.

8. PDF Export

InfoNinja can export summaries, research briefs, citation lists, or entire conversations as cleanly formatted PDFs with headers, bullet points, and page numbers.

9. Best Practices

  • Use clean URLs without tracking parameters.
  • Upload text-based PDFs when possible.
  • Break long research tasks into smaller prompts.
  • Ask for structured output.
  • Use follow-up questions to refine results.

10. Troubleshooting

  • URL didn’t load: The site may block scraping or require login.
  • PDF has no text: InfoNinja will OCR it for analysis but will not create a searchable PDF.
  • Citations missing page numbers: The file may not contain embedded page data.
  • PDF export looks empty: Request a specific summary or section first.
  • Answer seems incomplete: Smart Search may still be running; ask a follow-up question.

11. Data Storage Disclaimer (Under Active Development)

InfoNinja currently uses a combination of volatile in-memory chat buffers and a development-stage database layer that stores document chunks and recent context. This database storage is part of an early prototype design and is not intended for long-term or multi-user environments.

Work is underway to remove persistent storage and transition InfoNinja to a fully volatile, session-only memory model. Until then, InfoNinja should be considered in development and not used for sensitive or private documents.

12. Privacy and Data Handling

URLs are fetched temporarily for analysis, uploaded files are processed in-session, and no long-term storage occurs beyond the temporary development DB layer. No analytics or sharing of research content occurs.

13. Summary

InfoNinja analyzes URLs, documents, and text; generates citations; organizes findings; and exports clean PDFs. Smart Search Augmentation enhances every answer with optimized search queries and real web results. It can OCR image-based PDFs for analysis but does not modify or rebuild PDFs. A transition to a fully volatile memory model is underway.

Chat‑With‑Documents User Manual

Before You Begin

Chat‑With‑Documents requires a valid login to the Unlocking AI Launch Platform, a stable internet connection, and one or more documents you want to analyze. It provides a clean, fast way to upload documents and chat with their contents.

1. What Chat‑With‑Documents Does

Chat‑With‑Documents is a simple, multi‑document conversational workspace. It allows you to upload documents, extract text, ask questions, compare files, summarize content, and export your conversation. It is designed for lightweight document interaction without advanced research features.

2. How It Works

Step 1 — Upload documents: Upload a single document or multiple documents at once. Additional documents can be added later in the same session.

Step 2 — Text extraction: The app extracts text from PDFs with embedded text, text files, and markdown. Scanned PDFs must be OCR’d first using the PDF OCR Processor or InfoNinja.

Step 3 — Chat with your documents: Ask questions, request summaries, compare documents, or explore ideas across files. The AI responds using the extracted text from all uploaded documents.

Step 4 — Optional tools: Use Play, Auto TTS, Export PDF, or Print to enhance your workflow.

3. Multi‑Document Support

You can upload multiple documents into the same session. Chat‑With‑Documents extracts text from each file and makes all content available to the AI at once. This enables cross‑document comparison, combined summaries, conflict checks, and multi‑source analysis.

4. Supported Inputs

  • PDFs with embedded text
  • Text files (.txt)
  • Markdown (.md)
  • Multiple files at once
  • Additional uploads during the session

Unsupported:

  • Scanned/image‑only PDFs (no OCR)
  • Paywalled documents requiring login
  • Extremely large files that exceed browser limits

5. Using Chat‑With‑Documents

Upload documents: Drag and drop files or click the upload area.

Ask questions: Summaries, comparisons, explanations, or combined insights.

Use TTS: Play reads the current response; Auto TTS reads every response automatically.

Export or print: Export the conversation as a PDF or print directly.

6. Best Practices

  • Upload all documents you want to compare in the same session.
  • Break complex questions into smaller steps.
  • Ask for structured output.
  • Refresh the page to start a clean session.
  • OCR scanned PDFs before uploading.

7. Troubleshooting

  • PDF has no text: It may be a scanned PDF; OCR it first.
  • AI isn’t referencing all documents: Ensure all files uploaded successfully.
  • Export PDF is empty: Ask at least one question before exporting.
  • Documents disappeared: Refreshing the page clears the session.

8. Data Storage & Privacy

Chat‑With‑Documents uses only volatile, in‑session memory. Uploaded documents are processed in memory, extracted text is stored only for the duration of the session, and everything is cleared when the page is refreshed. No persistent storage, indexing, or cross‑user memory is used.

9. Summary

Chat‑With‑Documents is a lightweight, multi‑document chat tool that lets you upload files, extract text, ask questions, compare documents, and export your conversation. It does not perform OCR, does not use persistent storage, and keeps everything in temporary session memory.

To‑Do List User Manual

Before You Begin

The To‑Do List app is a lightweight, no‑friction task tracker built into the Unlocking AI Launch Platform. It requires a valid login, a modern browser, and uses no AI or backend database. All data is stored locally on your device.

1. What the To‑Do List App Does

This app provides a clean, minimal interface for managing tasks. You can add tasks, view your list, delete tasks when completed, and close the tab or browser and return later with your list intact.

2. How It Works

Add a task: Type a task into the input box and click Add. The task appears in your list with its own delete button.

Complete a task: Click the Delete button next to any task to remove it.

Persistence

Tasks are stored using localStorage, which means:

  • They persist across page refreshes
  • They persist if the tab is closed
  • They persist if the browser is closed
  • They persist after restarting your computer
  • They remain until you delete them or clear browser data

No server‑side storage or cloud sync is used.

3. Supported Features

  • Add unlimited tasks
  • Delete tasks individually
  • Persistent storage via localStorage
  • Clean, minimal UI
  • No backend dependencies
  • No AI dependencies

4. Limitations

  • Tasks do not sync across devices
  • Tasks do not sync across browsers
  • No categories, due dates, or reminders
  • No cloud storage
  • Clearing browser data removes all tasks

5. Best Practices

  • Keep task names short and clear
  • Delete tasks as soon as they’re completed
  • Avoid storing sensitive information (localStorage is not encrypted)
  • Use the list for lightweight, everyday tracking

6. Troubleshooting

  • Tasks disappeared: You may have cleared browser data or switched devices.
  • Add button doesn’t work: Ensure the input field contains text.
  • List didn’t update: Refresh the page — localStorage will reload your tasks.

7. Data Storage & Privacy

The To‑Do List app uses local browser storage only. There is no server database, no cloud sync, no cross‑user data, no analytics, and no AI processing. Your tasks stay on your device and nowhere else.

8. Summary

The To‑Do List app is a simple, persistent, local task tracker. Add tasks, delete them when done, and rely on localStorage to keep your list intact across refreshes, tab closures, and browser restarts. No AI, no backend, no complexity — just a clean, reliable checklist.

VoiceChat App User Manual

Before You Begin

The VoiceChat App enables real-time voice communication between up to five participants using secure, peer-to-peer WebRTC connections. You will need a valid login, a microphone, a modern browser with WebRTC support, and a Room ID and Passcode. This app does not use AI features.

1. What the VoiceChat App Does

The VoiceChat App provides real-time voice communication using direct, peer-to-peer audio transmission. Audio is encrypted end-to-end, transmitted directly between participants, and never processed or stored by any server. The system is designed for users who want simple, fast, and private voice communication without the overhead of traditional conferencing systems.

2. How It Works

Join a room: Enter a Room ID, Passcode, and Display Name, then click Join.

Establishing connections: Once inside the room, your microphone activates, a WebRTC RTCPeerConnection is created, ICE candidates are gathered, and the signaling server announces your presence so peers can connect.

Real-time audio: Audio is captured, encoded, encrypted, transmitted, played, and then discarded immediately. No audio is stored, logged, or processed by any server.

3. Multi-Peer Mesh Support

The VoiceChat App uses a mesh topology. Each participant connects directly to every other participant, and audio is exchanged peer-to-peer. Up to five participants are supported per room.

4. Privacy & Security Model

End-to-end encryption: All audio streams use DTLS-SRTP encryption with peer-negotiated keys. The server never has access to audio or encryption keys.

Ephemeral audio: Audio is captured, encrypted, sent, played, and discarded immediately. No recordings, logs, or transcripts are created.

Server-blind architecture: The signaling server only exchanges room metadata, SDP offers/answers, and ICE candidates. It never handles audio.

Design intent: The system is designed to support highly confidential conversations by ensuring no server-side audio, no storage, no logs, no transcripts, no analytics, and no cloud retention. This architecture makes monitoring technically difficult and operationally impractical while avoiding absolute claims.

5. Using the VoiceChat App

  • Join a room: Enter your Room ID, Passcode, and Display Name, then click Join.
  • Talk normally: Your microphone becomes active once connections form.
  • Leave the room: Closing the tab or navigating away disconnects you immediately.

6. Best Practices

  • Use headphones to avoid echo
  • Keep the participant count under five
  • Ensure microphone permissions are enabled
  • Use stable network connections for smoother audio
  • Use short, simple Room IDs and Passcodes for testing

7. Troubleshooting

  • Cannot join: Check the Room ID and Passcode.
  • No one can hear you: Verify microphone permissions.
  • Audio cuts out: Network instability or too many peers may be the cause.
  • Echo or feedback: Use headphones.
  • A peer disappeared: They may have closed their tab or lost connection.

8. Data Storage & Privacy

The VoiceChat App uses no server-side audio storage, no cloud recording, no transcripts, no analytics, and no persistent logs. Only minimal signaling metadata is exchanged to establish connections. All audio remains encrypted and peer-to-peer.

9. Summary

The VoiceChat App provides private, real-time, peer-to-peer voice communication using WebRTC mesh networking. Audio is encrypted end-to-end, transmitted directly between participants, and discarded immediately after playback. No servers process or store audio, making the system lightweight, fast, and designed for highly confidential conversations.

Platform App User Manual

Before You Begin

The Platform App is the core system that powers the entire Unlocking AI ecosystem. It provides authentication, subscription management, access control, and app hosting. All other apps run on top of the Platform App. This manual describes the Platform App itself — not the individual apps it hosts.

1. What the Platform App Does

The Platform App is the foundation of the entire system. It handles secure user authentication, subscription and billing logic, role-based access control, app routing, session management, and administrative oversight. It is the operating system for the Unlocking AI suite.

2. Core Responsibilities

Authentication: Manages account creation, login, logout, password resets, session tokens, and secure cookie handling.

Subscription Management: The subscription system is fully implemented and supports free and paid tiers, billing cycles, and premium app access. However, during the current open-access period, all apps are available for free and no subscription restrictions are enforced.

App Hosting & Routing: Exposes the full suite of apps through a unified dashboard with isolated routes and per-app access control.

Admin Interface: Allows administrators to view users, manage subscriptions, adjust access, and oversee system status.

3. Architecture Overview

User Accounts: Each user has a unique account, subscription tier, and set of accessible apps. No app stores its own user identity data; everything flows through the Platform App.

App Isolation: Each app is sandboxed, independently routed, independently permissioned, and stateless relative to other apps. The Platform App ensures no cross-app data leakage.

Security Model: Uses secure session tokens, encrypted transport (HTTPS), role-based access control, server-side validation, and strict routing rules.

4. Using the Platform App

  • Log in: Enter your credentials to authenticate.
  • Access the Dashboard: View the list of apps available to your account.
  • Launch an App: Click any app tile to open it. The Platform App verifies your access before loading the app.
  • Manage Your Account: Update your profile, change your password, view subscription status, and manage billing.
  • Log out: Ends your session and revokes access to all apps.

5. Best Practices

  • Keep your login credentials secure
  • Log out when using shared devices
  • Review your subscription tier once paid access becomes active
  • Use modern browsers for best compatibility
  • Contact an admin if access seems incorrect

6. Troubleshooting

  • Cannot log in: Check your credentials or reset your password.
  • App is locked: During the current free-access period, no apps should be locked. Contact an admin if one appears unavailable.
  • Session expired: Log in again; sessions expire for security.
  • Error opening an app: Your access token may be invalid; log out and back in.

7. Data Handling & Privacy

The Platform App stores account information, subscription status, and authentication tokens. It does not store app data, user content from other apps, audio, documents, or tasks. Each app handles its own data independently. The Platform App manages identity and access, not content.

8. Summary

The Platform App is the backbone of the Unlocking AI ecosystem. It manages authentication, subscriptions, access control, and app hosting. All other apps depend on it for secure, consistent access. The subscription system is fully implemented, but during the current open-access period, all apps remain free and no subscription tiers are enforced.