diff --git a/gitbooks/README.md b/gitbooks/README.md index 08504cffe..665e5052e 100644 --- a/gitbooks/README.md +++ b/gitbooks/README.md @@ -14,8 +14,6 @@ Every AI model in the world, all 200+ of them, shares the same fundamental limit OpenHuman solves this with three core innovations: -
- **Neocortex**, a human-like memory engine that can accurately work with over 1 billion tokens. Neocortex indexes 10 million tokens in under 10 seconds, costs $1 to index 5 million tokens, and runs on a MacBook Air CPU with zero LLM dependency. Neocortex goes beyond vector databases. It understands time, entities, and relationships, and it models memory the way the human brain does: by compressing, forgetting strategically, and building knowledge graphs. **A personalized subconscious system** inspired by the Purkinje cell, a specialized neuron in the human brain responsible for random thoughts and consciousness. OpenHuman's subconscious triggers over 10,000 memory recall loops per day for under $1, producing proactive insights, pattern recognition, and emergent "thoughts" that feed into a self-learning loop. diff --git a/gitbooks/SUMMARY.md b/gitbooks/SUMMARY.md index 1849e39b0..8880e0dca 100644 --- a/gitbooks/SUMMARY.md +++ b/gitbooks/SUMMARY.md @@ -2,33 +2,57 @@ ## Overview -- [Welcome to OpenHuman](README.md) -- [How It Works](overview/how-it-works.md) -- [Getting Started](overview/getting-started.md) +* [Welcome to OpenHuman](README.md) +* [How It Works](overview/how-it-works.md) +* [Getting Started](overview/getting-started.md) +* [App Navigation](overview/app-navigation.md) + +## FEATURES + +* [Screen Intelligence](features/screen-intelligence.md) +* [Inline Autocomplete](features/inline-autocomplete.md) +* [Voice Speech To Text](features/voice-speech-to-text.md) + +## SETTINGS REFERENCE + +* [Account & Security](settings-reference/account-and-security.md) +* [Automation & Channels](settings-reference/automation-and-channels.md) +* [AI & Skills](settings-reference/ai-and-skills.md) +* [Developer Options](settings-reference/developer-options.md) ## Technology -- [Neocortex](technology/neocortex.md) -- [The Subconscious](technology/the-subconscious.md) -- [Architecture](technology/architecture.md) +* [Neocortex](technology/neocortex.md) +* [The Subconscious](technology/the-subconscious.md) +* [Architecture](technology/architecture.md) ## Product -- [Platform & Availability](product/platform.md) -- [Skills & Integrations](product/skills-and-integrations.md) -- [Teams & Organizations](product/teams.md) -- [Privacy & Security](product/privacy-and-security.md) -- [Pricing](product/pricing.md) +* [Platform & Availability](product/platform.md) +* [Skills & Integrations](product/skills-and-integrations.md) +* [Teams & Organizations](product/teams.md) +* [Intelligence](product/intelligence.md) +* [Rewards](product/rewards.md) +* [Messaging Channels](product/messaging-channels.md) +* [Privacy & Security](product/privacy-and-security.md) +* [Pricing](product/pricing.md) ## Use Cases -- [Overview](use-cases/overview.md) +* [Overview](use-cases/overview.md) +* [Knowledge Workers](use-cases/knowledge-workers.md) +* [Individual Power Users](use-cases/individual-power-users.md) +* [Traders and Analysts](use-cases/traders-and-analysts.md) +* [DAO and Web3 Communities](use-cases/dao-and-web3-communities.md) +* [Community Managers & Moderators](use-cases/community-managers-and-moderators.md) +* [Distributed Teams](use-cases/distributed-teams.md) +* [Service Providers & Contributors](use-cases/service-providers-and-contributors.md) ## Resources -- [FAQ & Troubleshooting](resources/faq.md) +* [FAQ & Troubleshooting](resources/faq.md) ## Legal -- [Terms & Conditions](legal/terms-of-use.md) -- [Privacy Policy](legal/privacy-policy.md) +* [Terms & Conditions](legal/terms-of-use.md) +* [Privacy Policy](legal/privacy-policy.md) diff --git a/gitbooks/features/inline-autocomplete.md b/gitbooks/features/inline-autocomplete.md new file mode 100644 index 000000000..e03eae80b --- /dev/null +++ b/gitbooks/features/inline-autocomplete.md @@ -0,0 +1,49 @@ +--- +icon: rotate +--- + +# Inline Autocomplete + +Inline Autocomplete suggests text completions while you type in any application. Powered by your local AI model and your memory context. + +#### How It Works + +As you type, OpenHuman watches for pauses (controlled by debounce). After you pause, it generates a completion based on what you have typed plus relevant context from memory. Press Tab to accept. + +The more you accept suggestions, the more it learns your style through Personalization History. + +#### Runtime Info + +The settings page shows: platform support, enabled/running status, current phase, debounce timing, active model (e.g., gemma3:12b-it-q4\_K\_M), last error, and current suggestion. Start/Stop buttons available. + +#### Settings + +Navigate to Settings > Automation & Channels > Inline Autocomplete. + +**Enabled:** Master toggle. + +**Accept With Tab:** Accept suggestions by pressing Tab. On by default. + +**Debounce (ms):** Wait time after typing stops before generating a suggestion. Default: 120ms. + +**Max Chars:** Maximum characters per suggestion. Default: 384. + +**Style Preset:** Controls tone and style. Default: "Balanced." + +**Style Instructions:** Free-text field for your preferred writing style (e.g., "Write in short, direct sentences. Avoid formal language."). + +**Disabled Apps:** One bundle/app token per line. Autocomplete won't activate in listed apps. + +Click "Save Autocomplete Settings" after changes. + +#### Personalization History + +Tracks accepted completions to improve future suggestions. Clear History button resets personalization. Begins once you accept your first suggestion. + +#### Testing + +Test section with optional Context Override field. Buttons: Get Suggestion, Accept Suggestion, Debug Focus. + +#### Live Logs + +Real-time log showing autocomplete events with timestamps. Clear button to reset. diff --git a/gitbooks/features/screen-intelligence.md b/gitbooks/features/screen-intelligence.md new file mode 100644 index 000000000..7e1dedf20 --- /dev/null +++ b/gitbooks/features/screen-intelligence.md @@ -0,0 +1,57 @@ +--- +icon: desktop-arrow-down +--- + +# Screen Intelligence + +Screen Intelligence captures your screen, summarizes what you see, and feeds the context into OpenHuman's memory. It runs locally on your device. No screenshots or screen data are sent to any server. + +#### How It Works + +When a Screen Intelligence session is active, OpenHuman captures your screen at a configurable frame rate (default: 1 frame per second). Each capture is processed locally by the vision model, which generates a text summary of what is on screen. These summaries are ingested into your memory and become part of the context OpenHuman can reference when answering questions. + +The result: you can ask OpenHuman "What was I working on this morning?" or "What was on that dashboard I looked at earlier?" and get answers based on your actual screen activity. + +#### Setup + +Screen Intelligence requires three macOS permissions: + +**Screen Recording:** Allows OpenHuman to capture screen contents. Grant this in System Settings > Privacy & Security > Screen Recording. + +**Accessibility:** Allows OpenHuman to read window titles and application context. Grant this in System Settings > Privacy & Security > Accessibility. + +**Input Monitoring:** Allows OpenHuman to detect keyboard and mouse activity for context. Grant this in System Settings > Privacy & Security > Input Monitoring. + +The settings page shows the current status of each permission (GRANTED or DENIED) with buttons to request each one. After granting permissions in System Settings, click "Restart & Refresh Permissions" in OpenHuman so the app picks up the changes. + +#### Configuration + +**Enabled:** Master toggle to turn Screen Intelligence on or off. + +**Mode:** Controls which apps are captured. Default is "All Except Blacklist," which captures everything except apps on your denylist. You can switch to an allowlist mode to capture only specific apps. + +**Baseline FPS:** Frames per second for screen capture. Default is 1 (one capture per second). Lower values reduce resource usage, higher values capture more detail. + +**Keep Screenshots:** When enabled, saves captured screenshots to your workspace instead of deleting them after processing. Useful for debugging or review, but uses more storage. + +**Allowlist:** One rule per line. When in allowlist mode, only apps matching these rules are captured. + +**Denylist:** One rule per line. Apps matching these rules are never captured, regardless of mode. Default entries include: 1password, keychain, wallet. Add any sensitive applications here to ensure they are never captured. + +Click "Save Screen Intelligence Settings" after making changes. + +#### Session Controls + +**Start Session:** Begins a capture session. Screen Intelligence starts monitoring your screen according to your policy settings. + +**Stop Session:** Ends the current capture session. + +**Analyze Now:** Triggers immediate vision processing on the current screen without waiting for the next scheduled capture. + +**Panic Stop:** Press Cmd+Shift+. to immediately stop all capture. Useful if you navigate to sensitive content unexpectedly. + +The session panel shows: Status (Running/Stopped), Remaining time, Frames captured (ephemeral, not stored permanently unless Keep Screenshots is on), Vision status (idle/processing), Vision queue depth, and Last vision timestamp. + +#### Vision Summaries + +After processing, vision summaries appear in this section. Each summary is a text description of what was on screen at the time of capture. These summaries are what get ingested into your memory, not the raw screenshots. diff --git a/gitbooks/features/voice-speech-to-text.md b/gitbooks/features/voice-speech-to-text.md new file mode 100644 index 000000000..4c84280ed --- /dev/null +++ b/gitbooks/features/voice-speech-to-text.md @@ -0,0 +1,10 @@ +--- +icon: head-side-speak +--- + +# Voice Speech To Text + +Voice Speech To Text uses your microphone for dictation and voice-driven chat. All speech recognition runs locally on your device using the local AI model. No audio is sent to any server. + +Configure from the Skills tab (click "Open settings" next to Voice Speech To Text) or from Settings > Automation & Channels. + diff --git a/gitbooks/legal/privacy-policy.md b/gitbooks/legal/privacy-policy.md index 482eb06a3..a723c7601 100644 --- a/gitbooks/legal/privacy-policy.md +++ b/gitbooks/legal/privacy-policy.md @@ -1,3 +1,7 @@ +--- +icon: key +--- + # Privacy Policy Last Updated: 02/02/2026 @@ -18,9 +22,9 @@ We do not collect phone numbers, contact lists, precise location data, device id ## How We Use Information -We use information solely to operate, maintain, and provide the Service, including to: • Perform tasks explicitly requested by the user +We use information solely to operate, maintain, and provide the Service, including to: • Perform tasks explicitly requested by the user -• Provide contextual assistance across files, applications, or workflows +• Provide contextual assistance across files, applications, or workflows • Improve reliability and security of the Service @@ -40,11 +44,11 @@ Users may request deletion of their data at any time. Upon such request, all ass ## Legal Bases for Processing -Where required by law, we process personal data based on one or more of the following legal bases: +Where required by law, we process personal data based on one or more of the following legal bases: -• User consent +• User consent -• Performance of a contract +• Performance of a contract • Legitimate interests, where applicable and balanced against user rights diff --git a/gitbooks/legal/terms-of-use.md b/gitbooks/legal/terms-of-use.md index dfca842f0..526ba128b 100644 --- a/gitbooks/legal/terms-of-use.md +++ b/gitbooks/legal/terms-of-use.md @@ -1,3 +1,7 @@ +--- +icon: file-contract +--- + # Terms & Conditions Last Updated: 02/02/2026 @@ -10,13 +14,13 @@ The Service is a system-level AI assistant designed to help users complete tasks ## User Responsibilities -You agree to: +You agree to: -• Use the Service in compliance with applicable laws and regulations +• Use the Service in compliance with applicable laws and regulations -• Use the Service only with content and systems you have the right to access +• Use the Service only with content and systems you have the right to access -• Not use the Service for surveillance, harassment, or unlawful activity +• Not use the Service for surveillance, harassment, or unlawful activity • Not attempt to reverse engineer, interfere with, or misuse the Service diff --git a/gitbooks/overview/app-navigation.md b/gitbooks/overview/app-navigation.md new file mode 100644 index 000000000..8dbe4fbdd --- /dev/null +++ b/gitbooks/overview/app-navigation.md @@ -0,0 +1,32 @@ +--- +description: The OpenHuman desktop app has five main tabs in the bottom navigation bar. +icon: location-dot +--- + +# App Navigation + +#### Chat + +Your primary interaction surface. Talk to OpenHuman, ask questions, give instructions, and receive compressed intelligence from your connected data sources. Messages are processed by the local AI model running on your device. + +#### Skills + +View and configure OpenHuman's capabilities. Skills are organized in three groups: + +**Built-in Skills** are core desktop capabilities: Screen Intelligence, Text Auto-Complete, and Voice Speech To Text. Each has an "Open settings" button for configuration. + +**Channel Integrations** connect messaging platforms. Telegram and Discord are available, each requiring configuration before use. + +**3rd Party Skills** connect external data sources. Gmail and Notion are currently available. Each shows connection status, local storage size, and file count. + +#### Intelligence + +The window into OpenHuman's memory and reasoning. View your knowledge graph, memory statistics, ingestion activity, and extracted insights. Run analysis to extract actionable items from your connected skills. Includes sub-tabs for Memory (active), Subconscious (coming soon), and Dreams (coming soon). + +#### Rewards + +Earn community roles by using OpenHuman and participating in the Discord community. Track your progress across engagement milestones. + +#### Settings + +Configure your account, billing, AI model, automation features, messaging channels, and developer tools. Organized into four sections: Account & Security, Automation & Channels, AI & Skills, and Developer Options. diff --git a/gitbooks/overview/getting-started.md b/gitbooks/overview/getting-started.md index 4838eb06c..7d4a725b9 100644 --- a/gitbooks/overview/getting-started.md +++ b/gitbooks/overview/getting-started.md @@ -10,9 +10,33 @@ OpenHuman is open source under the GNU GPL3 license. The codebase is available a *** +#### System Requirements + +OpenHuman is a desktop application for macOS. Download it from [openhuman.ai](https://openhuman.ai). + +OpenHuman runs a local AI model on your machine. The experience scales with your hardware. + +| Model Tier | Download Size | Minimum RAM | GPU Required | Model | +| ------------------------- | ------------- | ----------- | ------------------------- | ---------------------- | +| Lightweight | \~1 GB | 4 GB | No | gemma3:1b-it-q4\_0 | +| Balanced | \~3 GB | 8 GB | No | gemma3:4b-it-qat | +| Performance (Recommended) | \~8 GB | 16 GB | Yes (Apple Silicon Metal) | gemma3:12b-it-q4\_K\_M | + +The app detects your system specs automatically and recommends the best tier. You can change the model tier at any time from Settings > AI & Skills > Local AI Model. + +#### macOS Permissions + +Some features require macOS permissions. You will be prompted during setup, but you can also configure them later in Settings > Automation & Channels. + +**Screen Intelligence** requires: Screen Recording, Accessibility, and Input Monitoring permissions. + +**Inline Autocomplete** requires: Accessibility and Input Monitoring permissions. + +After granting permissions in System Settings, return to OpenHuman and click "Restart & Refresh Permissions" so the app picks up the grants. + ## Download & Install -Download the OpenHuman desktop app from the official website at [tryopenhuman.com](https://tryopenhuman.com). The desktop app runs local AI models on your machine for private screen analysis, auto-complete, and chat. +Download the OpenHuman desktop app from the official website at [tryopenhuman.com](https://tryopenhuman.com). The desktop app runs local AI models on your machine for private screen analysis, inline autocomplete, and chat. OpenHuman runs natively on @@ -51,13 +75,13 @@ Screen Intelligence is one of OpenHuman's core capabilities. It captures your sc *** -### Enable Auto-complete +### Enable Inline Autocomplete -Auto-complete uses your Neocortex memory combined with the local model to suggest relevant text completions on any input surface across your system. +Inline autocomplete uses your Neocortex memory combined with the local model to suggest relevant text completions on any input surface across your system. -Because it draws on your accumulated context, suggestions reflect your actual terminology, projects, contacts, and patterns. This is different from generic auto-complete: it knows what you have been working on, who you have been talking to, and what is currently relevant. +Because it draws on your accumulated context, suggestions reflect your actual terminology, projects, contacts, and patterns. This is different from generic inline autocomplete: it knows what you have been working on, who you have been talking to, and what is currently relevant. -Auto-complete runs entirely on your local model. No keystroke data or text input is sent to any server. You can enable or disable it from settings at any time. +Inline autocomplete runs entirely on your local model. No keystroke data or text input is sent to any server. You can enable or disable it from settings at any time. *** diff --git a/gitbooks/overview/how-it-works.md b/gitbooks/overview/how-it-works.md index b0a076bf1..78a16d8fa 100644 --- a/gitbooks/overview/how-it-works.md +++ b/gitbooks/overview/how-it-works.md @@ -8,7 +8,7 @@ OpenHuman is built on three technology pillars: a memory engine called Neocortex ### The problem OpenHuman solves -Current AI systems fall into the category of Artificial Narrow Intelligence (ANI). They perform well within bounded domains, but they depend on carefully designed architectures and human-defined operational boundaries. They lack three things needed to get closer to general intelligence: persistent memory at scale, a form of consciousness or instinct, and the ability to ingest data without becoming inaccurate or expensive. +Current AI systems fall into the category of Artificial Narrow Intelligence (ANI). They perform well within bounded domains but they depend on carefully designed architectures and human-defined operational boundaries. They lack three things needed to get closer to general intelligence: persistent memory at scale, a form of consciousness or instinct, and the ability to ingest data without becoming inaccurate or expensive. Research from Google's Titans project demonstrates the core issue clearly. As LLMs try to absorb more data, they become less accurate. Larger context windows do not solve this. They make the models slower, more expensive, and more prone to errors. This means the path to superhuman AI requires solving for context that is high accuracy, high speed, and low cost. diff --git a/gitbooks/product/intelligence.md b/gitbooks/product/intelligence.md new file mode 100644 index 000000000..4a1bf5880 --- /dev/null +++ b/gitbooks/product/intelligence.md @@ -0,0 +1,54 @@ +--- +description: >- + The Intelligence tab shows what OpenHuman knows and how its memory is growing. + Open it from the bottom navigation bar. +icon: brain-circuit +--- + +# Intelligence + +### System Status + +The top of the page shows the current system state (System Idle, Analyzing, etc.) and an "Analyze Now" button. Clicking Analyze Now triggers immediate analysis of your connected skills to extract actionable items, insights, and memory updates. + +**Memory** (active): Your knowledge graph, statistics, and ingestion history. + +**Subconscious** (coming soon): Background reasoning loop that runs continuously between your prompts, synthesizing thoughts and detecting patterns. + +**Dreams** (coming soon): Deeper processing that consolidates and restructures memory during idle periods. + +### Memory + +The Memory section displays six key metrics: + +**Storage:** Total size of your local memory store, with file count. + +**Documents:** Number of documents ingested into memory. + +**Namespaces:** Organizational groupings within your knowledge graph. Different data sources and contexts are separated into namespaces. + +**Relations:** Entity relationships extracted and stored in the knowledge graph. These connect people, projects, decisions, and concepts across your data. + +**First Memory:** When your first memory was created, with the timestamp of the most recent memory. + +**Sessions:** Number of active sessions and total tokens processed. + +### Memory Graph + +A visual representation of entities and their relationships. This populates as you ingest more data from connected sources. The graph shows how people, projects, topics, and decisions connect across your data. + +### Intelligent Insights + +Extracted facts, preferences, and relationships from ingested documents. Populates after you run an analysis. Insights include inferred preferences, detected patterns, and factual information about entities in your data. + +### Ingestion Activity + +A heatmap showing ingestion events over time, similar to a GitHub contribution graph. Displays date range, total events, and peak activity. Helps you understand how actively OpenHuman is processing your data. + +### Files & Management + +Shows total count of files, namespaces, and documents in your memory store. Expandable for file-level inspection and management. + +### Analysis + +A search bar lets you search for actionable items across all sources with source filtering. The "Analyze Now" button extracts actionable items from your connected skills. Results appear below, organized by source. diff --git a/gitbooks/product/messaging-channels.md b/gitbooks/product/messaging-channels.md new file mode 100644 index 000000000..21e838771 --- /dev/null +++ b/gitbooks/product/messaging-channels.md @@ -0,0 +1,33 @@ +--- +description: >- + OpenHuman communicates with you through messaging channels. You choose your + default, and OpenHuman sends proactive messages, briefings, and responses + there. +icon: message-plus +--- + +# Messaging Channels + +#### Available Channels + +**Telegram:** The primary messaging channel. Connect your Telegram account and OpenHuman can send and receive messages, manage chats, search history, create groups, and perform 80+ actions on your behalf. All actions run through your own encrypted credentials. + +**Discord:** Send and receive messages via Discord. Connect your account to receive OpenHuman messages in Discord. + +**Web:** A browser-based chat interface within the OpenHuman desktop app. Messages stay entirely local. + +#### Setting Your Default Channel + +Navigate to Settings > Automation & Channels > Messaging Channels. Select your preferred default (Telegram, Discord, or Web). The active route status shows which channel is currently in use. + +#### Configuring a Channel + +Click the channel name in Channel Integrations to open its configuration page. + +**Telegram** offers two credential modes: connect via OpenHuman (one-click, encrypted) or provide your own credentials for maximum control. + +**Discord** follows a similar authorization pattern. + +#### Telegram Capabilities + +When connected, OpenHuman has access to 80+ Telegram actions. See the T[elegram Capabilities Reference](messaging-channels.md#telegram-capabilities) for the complete list. diff --git a/gitbooks/product/platform.md b/gitbooks/product/platform.md index 0ace7c93a..940836fd1 100644 --- a/gitbooks/product/platform.md +++ b/gitbooks/product/platform.md @@ -31,7 +31,7 @@ The desktop app is the primary OpenHuman experience. It runs local AI models on **Screen Intelligence** captures your screen approximately every 5 seconds, processes captures through the on-device vision model, and produces structured context summaries. This feature requires the desktop app because it needs access to your screen and local compute for real-time processing. Per-app permissions let you control which applications are monitored. -**Auto-complete** uses your Neocortex memory context combined with the local model to generate relevant text completions on any input surface. This runs entirely on-device and is available on desktop platforms. +**Inline Autocomplete** uses your Neocortex memory context combined with the local model to generate relevant text completions on any input surface. This runs entirely on-device and is available on desktop platforms. **Local model inference** for chat, vision, speech-to-text, and text-to-speech runs on your desktop hardware. No GPU is required, though GPU acceleration is used when available. @@ -55,7 +55,7 @@ OpenHuman is built as a native application rather than a web wrapper for three r OpenHuman operates across three layers: -**Application layer.** The native app on your device handles the interface, user input, local state, credential management, local model inference, Screen Intelligence, Auto-complete, and skill execution. This layer is responsible for everything you see and interact with, and for all on-device AI processing. +**Application layer.** The native app on your device handles the interface, user input, local state, credential management, local model inference, Screen Intelligence, **Inline Autocomplete**, and skill execution. This layer is responsible for everything you see and interact with, and for all on-device AI processing. **Intelligence layer.** OpenHuman's analysis, coordination, and intelligence systems run as a secure backend service. When a request requires deeper language processing, it is handled here. This layer is operated and maintained by OpenHuman. @@ -79,6 +79,6 @@ The connection is designed for resilience. If the network drops, OpenHuman recon OpenHuman's local state persists on your device. Your preferences, settings, and connected source configurations remain available even when you are offline. -On desktop, local model capabilities continue to function offline. Screen Intelligence continues capturing and processing locally. Auto-complete continues generating suggestions from cached memory context. Chat with the local model works without a network connection. +On desktop, local model capabilities continue to function offline. Screen Intelligence continues capturing and processing locally. **Inline Autocomplete** continues generating suggestions from cached memory context. Chat with the local model works without a network connection. Full analysis and intelligence features that require the server-side intelligence layer need a network connection. When connectivity is restored, the app resumes normal operation without requiring you to re-authenticate or reconfigure. diff --git a/gitbooks/product/pricing.md b/gitbooks/product/pricing.md index dd1b6fd55..5408416a1 100644 --- a/gitbooks/product/pricing.md +++ b/gitbooks/product/pricing.md @@ -4,44 +4,54 @@ icon: sack-dollar # Pricing -OpenHuman is currently in early alpha. During this phase, access is free for early users. +OpenHuman offers a free tier with optional paid upgrades for heavier usage. -*** +#### Plans -### Alpha Access +**Free ($0/month)** Base access to integrations and inference. Pay-as-you-go top-ups when included usage runs out. No subscription discount on premium usage. -All features are available at no cost during the alpha period, including Screen Intelligence, Auto-complete, Telegram integration, Notion and Google Sheets exports, and subconscious intelligence. +| Included monthly value | 7-day cycle budget | 5-hour cap | Discount | Storage | +| ---------------------- | ------------------ | ---------- | -------- | ------- | +| $1 | $0.50 | $0.15 | 0% | 100 MB | -Early users who join during alpha will receive advance notice before paid plans take effect, along with a transition period and loyalty benefits. +**Basic ($20/month)** Higher included premium usage every billing cycle. 20% premium-usage discount across integrations, bandwidth, and inference. Pay-as-you-go top-ups for overflow usage. -*** +| Included monthly value | 7-day cycle budget | 5-hour cap | Discount | Storage | +| ---------------------- | ------------------ | ---------- | -------- | ------- | +| $20 | $10 | $3 | 20% | 10 GB | -### Planned Pricing Model +**Pro ($200/month)** Largest included premium usage allocation. 40% premium-usage discount across integrations and inference. Best fit for heavy bandwidth and agent workloads. -OpenHuman will be priced as a subscription for individuals and teams. +| Included monthly value | 7-day cycle budget | 5-hour cap | Discount | Storage | +| ---------------------- | ------------------ | ---------- | -------- | ------- | +| $200 | $100 | $30 | 40% | 200 GB | -**Free tier:** Core functionality with local models. Screen Intelligence, Auto-complete, and basic Neocortex memory using on-device processing. +Monthly and Annual billing options are available. Annual billing lowers the effective monthly price. -**Paid subscription:** Access to premium cloud models for deeper analysis, full Neocortex intelligence at scale, unlimited subconscious processing, priority support, and advanced integrations. +#### Pay as You Go -Pricing details and tier breakdowns will be published before the alpha period ends. +Buy credits as you need them with no subscription required. If you have a subscription, your included budget is consumed first. -*** +Top-up options: +$5, +$10, +$25. -### Team Plans +**Coupon codes:** Enter a coupon code (format: XXXX-XXXX) in Settings > Account & Security > Billing & Usage to redeem credits. -Team plans will support shared coordination environments where multiple members need access under a single workspace. Team plans will include shared workflows, community-level intelligence, coordination support, and centralized billing. +#### Pay with Crypto -*** +You can choose to pay annually using crypto. Toggle this option in Settings > Account & Security > Billing & Usage. -### Credit System +#### Inference Budget -OpenHuman uses a credit-based system for usage-metered features beyond the subscription base. Credits are consumed when you use analysis, intelligence, and processing features. Credits can be earned through the referral program, in addition to purchase. +Your billing dashboard tracks usage within each cycle: daily spending, tokens consumed, 5-hour cap usage, and cycle end date. A progress bar shows remaining budget. -*** +#### Auto-Recharge -### Payment Methods +Toggle auto-recharge to automatically top up credits when your balance runs low. Requires a saved payment method (add a card in Settings > Account & Security > Billing & Usage). -**Card payments:** Standard subscription billing with the ability to upgrade, downgrade, or cancel at any time. +#### Why Upgrade? -**Cryptocurrency:** Alternative payment path for users who prefer crypto. +Higher tiers increase your premium-usage discount and included usage every cycle. Annual billing lowers the effective monthly price, and top-ups let you keep going when usage spikes. + +#### Where to Manage + +All billing, credits, and payment methods are managed in Settings > Account & Security > Billing & Usage. diff --git a/gitbooks/product/privacy-and-security.md b/gitbooks/product/privacy-and-security.md index 36381789a..1937fc4e2 100644 --- a/gitbooks/product/privacy-and-security.md +++ b/gitbooks/product/privacy-and-security.md @@ -18,7 +18,7 @@ OpenHuman operates on a principle of **zero retention** for message content. Whe **OS-level credential storage.** On desktop platforms, OpenHuman uses your operating system's secure keychain to store credentials and sensitive tokens. Credentials are never stored in plain text, browser storage, or application-level databases. -**On-device where possible.** Screen Intelligence, Auto-complete, local chat, voice processing, and encryption all happen on your device. Only tasks requiring deeper language processing beyond local model capabilities are handled server-side, under the same privacy constraints. +**On-device where possible.** Screen Intelligence, Inline autocomplete, local chat, voice processing, and encryption all happen on your device. Only tasks requiring deeper language processing beyond local model capabilities are handled server-side, under the same privacy constraints. ### Screen Intelligence Privacy @@ -36,9 +36,9 @@ Screen Intelligence captures your screen approximately every 5 seconds and proce *** -### Auto-complete Privacy +### Inline autocomplete Privacy -Auto-complete runs on the local model using cached memory context from Neocortex. No keystroke data, text input, or completion suggestions are sent to any server. The entire feature operates on-device. +Inline autocomplete runs on the local model using cached memory context from Neocortex. No keystroke data, text input, or completion suggestions are sent to any server. The entire feature operates on-device. *** diff --git a/gitbooks/product/rewards.md b/gitbooks/product/rewards.md new file mode 100644 index 000000000..461f53aec --- /dev/null +++ b/gitbooks/product/rewards.md @@ -0,0 +1,32 @@ +--- +description: The Rewards tab tracks your engagement with OpenHuman and the community. +icon: hand-holding-star +--- + +# Rewards + +#### Discord Rewards + +Join the OpenHuman Discord, connect your account, and unlock roles as you use the app. Two buttons at the top: Join Discord (opens the invite) and Connect Discord (links your Discord account). + +#### Progress + +A progress tracker shows roles unlocked out of total (e.g., 2/5) with a progress bar. + +Stats displayed: Discord linked status (Yes/No), total messages sent, and current plan (Free, Basic, or Pro). + +#### Roles + +Roles unlock based on your activity: + +**First Contact:** Send your first message to OpenHuman. Unlock action: Start one chat. + +**Discord Pilot:** Link Discord messaging so OpenHuman can reach you there. Unlock action: Connect Discord in the Messaging settings. + +**Power User:** Build momentum by actively using OpenHuman across multiple sessions. Unlock action: Reach 10 total chat messages. + +**Supporter:** Unlock with an active paid plan. Unlock action: Upgrade to Basic or Pro. + +**Community Builder:** Bring another human into the network through the invite system. Unlock action: Redeem or participate in invites. + +Each role shows its status (Unlocked with a green checkmark, or Locked with a clock icon), a description, and the specific unlock action with current progress. diff --git a/gitbooks/product/skills-and-integrations.md b/gitbooks/product/skills-and-integrations.md index 86a362cd3..d505dc83d 100644 --- a/gitbooks/product/skills-and-integrations.md +++ b/gitbooks/product/skills-and-integrations.md @@ -4,88 +4,53 @@ icon: screwdriver # Skills & Integrations -OpenHuman extends its capabilities through two systems: a **skills engine** that adds new functionality to the assistant, and **integrations** that connect OpenHuman to external tools. Together, they allow OpenHuman to do more than analyze conversations; it can fetch data, run scheduled tasks, produce structured outputs, and write results to the tools you already use. +### Skills & Integrations ---- +OpenHuman's capabilities are organized into three categories, all accessible from the Skills tab in the bottom navigation. -## Skills Engine +#### Built-in Skills -Skills are sandboxed modules that run inside OpenHuman's native application. Each skill is an isolated unit with its own execution environment, storage, and permissions. Skills extend what the assistant can do without requiring an app update. +Core desktop capabilities that run locally on your device. -When you make a request, OpenHuman's AI automatically discovers which skills are available and decides whether to invoke them based on what you asked for. You do not need to manage skills manually. They are loaded, registered, and made available to the AI system transparently. +**Screen Intelligence:** Captures your screen, summarizes what is on it, and feeds useful context into memory. Configurable capture frequency (default: 1 frame per second), per-app allowlist and denylist controls, and privacy-first local processing. Configure in Settings > Automation & Channels > Screen Intelligence. -### What Skills Can Do +**Text Auto-Complete (Inline Autocomplete):** Suggests inline completions while you type in any application. Powered by your local AI model and your memory context. Accept suggestions with Tab. Configurable debounce timing, max characters, style presets, and per-app disable list. Configure in Settings > Automation & Channels > Inline Autocomplete. -Skills operate within defined boundaries. A skill can: +**Voice Speech To Text:** Uses the microphone for dictation and voice-driven chat with local speech recognition. All processing happens on-device. -- **Fetch external data:** make HTTP requests to APIs, retrieve information from services, and bring data into OpenHuman's context -- **Run on a schedule:** execute at defined intervals using cron-style scheduling, enabling automated monitoring, reporting, or data collection -- **Process and transform:** analyze, filter, or restructure data within its sandboxed environment, using its own local database for persistence -- **Write outputs:** produce structured results that the AI can present to you or export through integrations -- **Respond to events:** react to real-time events from the server, enabling workflows that trigger based on external conditions +#### Channel Integrations -### Isolation and Safety +Messaging platforms that OpenHuman can send and receive messages through. -Each skill runs in its own sandboxed environment with enforced resource limits. Skills cannot access each other's data or execution context. This means a misbehaving skill cannot affect the rest of the application or compromise your data. +**Telegram:** Full messaging integration with 80+ capabilities including sending, replying, forwarding, editing, and deleting messages, managing groups and channels, handling contacts, admin actions, and more. See the Telegram Capabilities Reference for the complete list. Two connection modes: connect via OpenHuman (one-click, encrypted) or provide your own credentials. -Skills are discovered from a curated registry and filtered by platform compatibility. Only skills that support your current platform are loaded. +**Discord:** Send and receive messages via Discord. Configuration required. -{% hint style="info" %} -Skills extend OpenHuman's capabilities without requiring trust beyond their declared scope. Each one is isolated, resource-limited, and independently revocable. -{% endhint %} +Both channels show connection status and can be configured by clicking the Configure button in the Skills tab. ---- +#### 3rd Party Skills -## Integrations +External data sources that OpenHuman ingests and reasons over. -Integrations connect OpenHuman to external tools so structured outputs can live where your team already works. Integrations are not about importing data into OpenHuman, they are about exporting intelligence out of it. +**Gmail:** Connects via Google API for comprehensive email integration. Once connected, OpenHuman syncs your email on a configurable interval (default: every 15 minutes). Shows local storage used and file count. To connect: go to Skills tab > 3rd Party Skills > Gmail > Configure. -### Supported Integrations +**Notion:** Workspace integration with 25 tools for pages, databases, and documents. Syncs on a configurable interval (default: every 20 minutes). Shows local storage used and file count. -**Telegram:** OpenHuman connects to your Telegram account to read and analyze conversations, extract signals, summarize discussions, and generate context-aware responses. Access is scoped to the conversations you reference in your requests. +To connect: go to Skills tab > 3rd Party Skills > Notion > Configure. -**Notion:** Export summaries, action items, decisions, and workflow records to your Notion workspace. OpenHuman writes structured documents that preserve the context of where outputs originated. Useful for building shared knowledge bases or living records. +**Important:** Gmail and Notion are available now as 3rd Party Skills. The Connections page in **Settings > Account & Security** shows deeper native integrations (Google, Notion, Web3 Wallet, Crypto Trading Exchanges) that are coming soon. To connect Gmail or Notion today, use the Skills tab, not the Connections page. -**Google Sheets:** Export tabular data such as logs, reports, extracted metrics, or tracking records. Useful for lightweight analysis, reporting, and audit trails. +#### Sync Intervals -These integrations are optional and can be enabled or disabled at any time from your settings. +Each connected 3rd party skill syncs on a configurable schedule. You can adjust intervals in **Settings > Automation & Channels > Cron Jobs**. -### How Data Moves +| Skill | Default Sync Interval | +| ------ | --------------------- | +| Gmail | Every 15 minutes | +| Notion | Every 20 minutes | -Data moves from OpenHuman to external tools **only when you explicitly request it**. +More frequent syncing keeps data fresher but uses more inference budget. -When a workflow, summary, or analysis is exported, OpenHuman writes structured data, not raw conversation text. There is no continuous sync, no background polling, and no automatic data transfer. Each export is a deliberate action you initiate. +#### Coming Soon -Integrations are write-oriented by default. OpenHuman does not read from external tools unless a specific workflow requires it and you have approved that behavior. - -{% hint style="info" %} -**You control what leaves OpenHuman.** Nothing is exported to external tools without a clear, user-initiated action. -{% endhint %} - ---- - -## From Conversation to Action - -One of OpenHuman's core capabilities is turning unstructured conversations into structured, actionable outputs. - -When discussions lead to decisions, commitments, or tasks, OpenHuman can recognize these moments (when you ask it to analyze a conversation) and represent them as **workflows as** structured objects that capture what needs to happen, who is involved, and where the decision came from. - -Workflows link back to the conversations they originated from, preserving intent and rationale. When connected to tools like Notion or Google Sheets, workflows can be exported without duplicating entire conversation histories. - -Workflows exist to reduce coordination friction, not to replace dedicated project management systems. OpenHuman does not auto-assign tasks, enforce deadlines, or create workflows without your input. - ---- - -## Intelligence Concepts - -OpenHuman works by translating raw conversation activity into a small number of structured concepts. These concepts allow the system to reason about communication, coordination, and trust without storing message histories. - -**Signals:** Structured interpretations extracted from conversations. A signal might represent a decision, warning, important update, or emerging theme. Signals are independent objects that can be filtered, referenced, or aggregated. - -**Claims:** Specific assertions that can be evaluated over time, such as predictions, commitments, or verifiable statements. Claims are tracked structurally and can resolve as accurate, inaccurate, or inconclusive. - -**Contributions:** Meaningful actions taken by participants within conversations, inferred from behavior and context. Contribution records support community recognition and value attribution. - -**Workflows:** Structured processes that originate from conversation, preserving the context of decisions and commitments as actionable objects. - -These concepts allow OpenHuman to support trust, coordination, and intelligence without becoming invasive or requiring the storage of raw conversation data. +Connections for Google (calendar, contacts), Notion (deeper native integration), Web3 Wallet, and Crypto Trading Exchanges are in development. These will appear in Settings > Account & Security > Connections when available. diff --git a/gitbooks/product/teams.md b/gitbooks/product/teams.md index 0f950be75..41979726e 100644 --- a/gitbooks/product/teams.md +++ b/gitbooks/product/teams.md @@ -8,19 +8,19 @@ OpenHuman supports team workspaces for organizations that need shared access, ce Teams are designed for professional environments where multiple people rely on OpenHuman and need shared administrative control without compromising individual privacy. ---- +*** ## Team Workspaces A team workspace is a shared environment where members access OpenHuman under a unified account structure. Workspaces provide: -- **Centralized billing:** one subscription covers all team members, with usage tracked per member -- **Member management:** invite, remove, and manage team members from a single dashboard -- **Shared configuration:** team-level settings and preferences that apply across the workspace +* **Centralized billing:** one subscription covers all team members, with usage tracked per member +* **Member management:** invite, remove, and manage team members from a single dashboard +* **Shared configuration:** team-level settings and preferences that apply across the workspace Each team has a unique identifier and can be managed independently from individual accounts. Users can belong to a team while still maintaining their personal OpenHuman account and settings. ---- +*** ## Roles and Permissions @@ -32,7 +32,7 @@ Team workspaces support three roles: **Member:** Standard access to OpenHuman within the team workspace. Members use the product normally and benefit from the team subscription, but cannot modify team settings or manage other members. ---- +*** ## Shared Intelligence @@ -50,15 +50,27 @@ All shared intelligence is derived from aggregated and anonymized signals. OpenH **Privacy is preserved in teams.** Shared intelligence is always aggregated. Individual message content and personal analysis remain private to each user. {% endhint %} ---- +*** ## Referral Program OpenHuman offers a referral system where existing users can invite others and earn credits. -- Generate a personal invite code from your account settings -- Share the code with colleagues or collaborators -- Both the referrer and the new user receive credits when the invitation is accepted -- Additional credit bonuses may apply when referred users subscribe to a paid plan +* Generate a personal invite code from your account settings +* Share the code with colleagues or collaborators +* Both the referrer and the new user receive credits when the invitation is accepted +* Additional credit bonuses may apply when referred users subscribe to a paid plan Referral credits can be used toward OpenHuman usage across any plan tier. + +*** + +## Managing Teams + +Navigate to Settings > Account & Security > Team to manage your teams. + +**Your Teams:** Shows all teams you belong to with your role (Owner or Member), plan tier (Free, Basic, or Pro), and status (Active or Inactive). + +**Create New Team:** Enter a team name and click Create. + +**Join Existing Team:** Enter an invite code provided by a team owner and click Join. diff --git a/gitbooks/resources/faq.md b/gitbooks/resources/faq.md index 4cd3dd82d..b5347737c 100644 --- a/gitbooks/resources/faq.md +++ b/gitbooks/resources/faq.md @@ -86,7 +86,7 @@ Messages are processed only to fulfill your request. They are **not permanently #### Can OpenHuman send messages on my behalf? -OpenHuman can send messages, reply, forward, and take other actions in your Telegram account, but only when you explicitly instruct it to. It never sends messages autonomously or acts without your direct request. +OpenHuman can send messages, manage chats, and perform 80+ actions through your connected Telegram account. All actions use your own encrypted credentials and run locally. You control which capabilities are active through the Skills settings *** @@ -208,3 +208,27 @@ Previously exported outputs (such as summaries written to Notion or Google Sheet If an issue persists after refining scope, checking permissions, and retrying the request, support may be required. Support is intended for system-level issues, not for disputing interpretations or outcomes. OpenHuman does not adjudicate trust or risk disagreements. + +*** + +#### **What are the system requirements?** + +OpenHuman requires macOS. The Lightweight model tier works on machines with 4 GB+ RAM. The recommended Performance tier requires 16 GB+ RAM and Apple Silicon. The app detects your hardware and recommends the best tier automatically. + +*** + +#### **How often does OpenHuman sync my email and Notion?** + +Gmail syncs every 15 minutes and Notion every 20 minutes by default. You can adjust these intervals in Settings > Automation & Channels > Cron Jobs. + +*** + +#### **What is the panic stop shortcut?** + +Press Cmd+Shift+. to immediately stop all Screen Intelligence capture. Useful if you navigate to sensitive content unexpectedly. + +*** + +#### **What apps are excluded from Screen Intelligence by default?** + +1password, keychain, and wallet are in the default denylist. You can add more apps in Settings > Automation & Channels > Screen Intelligence under the Denylist field. diff --git a/gitbooks/settings-reference/account-and-security.md b/gitbooks/settings-reference/account-and-security.md new file mode 100644 index 000000000..0b2964b70 --- /dev/null +++ b/gitbooks/settings-reference/account-and-security.md @@ -0,0 +1,41 @@ +--- +icon: user-lock +--- + +# Account & Security + +#### Billing & Usage + +View and manage your subscription, credits, and payment methods. See the Pricing page for plan details. + +Your billing dashboard shows: current plan, included monthly value, cycle budget, 5-hour cap, discount percentage, and storage. Below that, your inference budget tracks daily usage, tokens consumed this cycle, 5-hour cap spending, and cycle end date with a progress bar. + +**Pay as You Go:** Top up credits in +$5, +$10, or +$25 increments. Redeem coupon codes (XXXX-XXXX format). + +**Pay with Crypto:** Toggle for annual crypto payments. + +**Auto-Recharge:** Toggle to auto-top-up when balance is low. Requires a saved payment method. + +**Payment Methods:** Add a card to enable auto-recharge and plan upgrades. + +#### Recovery Phrase + +Manage your BIP39 recovery phrase for encryption and wallet access. This phrase is the key to your encrypted data. Store it securely offline. If you lose your recovery phrase, your encrypted data cannot be recovered. + +#### Team + +Manage your teams, members, and invites. + +**Your Teams:** Shows teams with role (Owner/Member), plan tier, and status. + +**Create New Team:** Enter a name and click Create. + +**Join Existing Team:** Enter an invite code and click Join. + +#### Connections + +Review and manage linked account connections. Currently showing as Coming Soon: Google (emails, contacts, calendar), Notion (tasks, documents), Web3 Wallet (trading), Crypto Trading Exchanges (trades with deep insights). + +Note: Gmail and Notion are available now as 3rd Party Skills in the Skills tab. The Connections page will offer deeper native integrations when they launch. + +**Privacy notice:** All data and credentials are stored locally with zero-data retention policy. Your information is encrypted and never shared with third parties. diff --git a/gitbooks/settings-reference/ai-and-skills.md b/gitbooks/settings-reference/ai-and-skills.md new file mode 100644 index 000000000..93e0be248 --- /dev/null +++ b/gitbooks/settings-reference/ai-and-skills.md @@ -0,0 +1,45 @@ +--- +icon: head-side-circuit +--- + +# AI & Skills + +#### Local AI Model + +Choose your model tier based on device capability. + +OpenHuman detects your system specs automatically: RAM, CPU cores, and GPU type. + +| Tier | Size | RAM | GPU | Model | Best For | +| ----------- | ------ | ------ | ------------------------- | ---------------------- | ------------------------------------ | +| Lightweight | \~1 GB | 4 GB+ | No | gemma3:1b-it-q4\_0 | Older machines, minimal resource use | +| Balanced | \~3 GB | 8 GB+ | No | gemma3:4b-it-qat | Moderate hardware | +| Performance | \~8 GB | 16 GB+ | Yes (Apple Silicon Metal) | gemma3:12b-it-q4\_K\_M | Best quality, recommended | + +The active tier is marked "ACTIVE" and the recommended tier is marked "RECOMMENDED." Switching tiers triggers a new model download. + +"Show Advanced" expands additional model configuration options. + +#### AI Configuration + +Configure persona, prompting behavior, and AI runtime settings. + +**AI System Overview:** Prompt and markdown orchestration handled in Rust runtime. Shows configuration status (Ready or Fallback Mode) and loading duration. + +**Local Model Runtime:** State (ready/loading/error) and target model. "Open Manager" and "Retry Download" buttons. + +**SOUL Persona Configuration:** Defines OpenHuman's identity and behavior. Shows identity name (OpenHuman), role (AI assistant), source, and load timestamp. "Refresh SOUL" reloads persona configuration. + +**TOOLS Configuration:** Available tools count and active skills count. "Refresh TOOLS" reloads tool definitions. + +**Refresh All AI Configuration:** Reloads the entire AI configuration stack. + +#### Skills + +Configure browser access and installed skill capabilities. + +**Browser Access:** Controls whether OpenHuman's browser tool can visit public domains. Private and file URLs are always blocked. + +"Restrict to allowlist" limits access to only approved domains. When unchecked, any public domain is accessible. + +Shows registered integrations list. diff --git a/gitbooks/settings-reference/automation-and-channels.md b/gitbooks/settings-reference/automation-and-channels.md new file mode 100644 index 000000000..ec05dc3b7 --- /dev/null +++ b/gitbooks/settings-reference/automation-and-channels.md @@ -0,0 +1,54 @@ +--- +icon: nfc-signal +--- + +# Automation & Channels + +### Automation & Channels + +Navigate to Settings > Automation & Channels. Configure desktop automation, screen capture, messaging, autocomplete, and scheduled jobs. + +#### Accessibility Automation + +Desktop permissions, assisted controls, and safety-bound sessions. + +**Permissions:** Accessibility (GRANTED/DENIED) and Input Monitoring (GRANTED/DENIED). Buttons: Request Accessibility, Open Input Monitoring, Refresh Status. + +**Features:** Three toggles: Screen Monitoring (observe screen for context), Device Control (interact with UI elements), Predictive Input (provide input predictions). + +**Session:** Status (Running/Stopped), remaining time, frames captured (ephemeral), panic stop shortcut (Cmd+Shift+.). Start Session, Stop Session, and Analyze Now buttons. + +**Vision Summaries:** Processed summaries from the most recent session. + +#### Screen Intelligence + +Window capture policy, vision summaries, and memory ingestion. See the full [Screen Intelligence](../features/screen-intelligence.md) page for detailed setup. + +#### Inline Autocomplete + +Predictive text style, app filters, and live completion controls. See the full [Inline Autocomplete](../features/inline-autocomplete.md) page for detailed setup. + +#### Messaging Channels + +Configure default messaging channel and auth modes. + +**Default Messaging Channel:** Choose between Telegram, Discord, or Web. Shows active route status. + +**Channel Integrations:** Configure auth modes for Telegram and Discord. Click channel name to open its configuration page. + +#### Cron Jobs + +Scheduled jobs that keep your data sources in sync. + +**Core Cron Jobs:** System-level jobs in the OpenHuman core scheduler database. + +**Runtime Skill Cron Settings:** Per-skill sync intervals. + +| Skill | Default Sync Interval | +| ------ | --------------------- | +| Gmail | Every 15 minutes | +| Notion | Every 20 minutes | + +Adjust intervals using the dropdown next to each skill. More frequent syncing keeps data fresher but uses more inference budget. + +Click "Refresh Cron Jobs" to reload the schedule. diff --git a/gitbooks/settings-reference/developer-options.md b/gitbooks/settings-reference/developer-options.md new file mode 100644 index 000000000..0110f5157 --- /dev/null +++ b/gitbooks/settings-reference/developer-options.md @@ -0,0 +1,6 @@ +--- +hidden: true +--- + +# Developer Options + diff --git a/gitbooks/technology/architecture.md b/gitbooks/technology/architecture.md index d1e0d372b..08b016533 100644 --- a/gitbooks/technology/architecture.md +++ b/gitbooks/technology/architecture.md @@ -1,3 +1,7 @@ +--- +icon: code-branch +--- + # Architecture OpenHuman is built on the OpenClaw architecture and open-sourced under the GNU GPL3 license. This page explains how the major components connect. diff --git a/gitbooks/technology/neocortex.md b/gitbooks/technology/neocortex.md index 75174ba32..91460e92b 100644 --- a/gitbooks/technology/neocortex.md +++ b/gitbooks/technology/neocortex.md @@ -1,3 +1,7 @@ +--- +icon: head-side-brain +--- + # Neocortex Neocortex is OpenHuman's memory engine. It is a human-like AI memory system designed to work accurately with over 1 billion tokens of data while supporting the computational demands of a subconscious system. diff --git a/gitbooks/technology/the-subconscious.md b/gitbooks/technology/the-subconscious.md index 12c122cbd..4107571bb 100644 --- a/gitbooks/technology/the-subconscious.md +++ b/gitbooks/technology/the-subconscious.md @@ -1,3 +1,7 @@ +--- +icon: head-side-gear +--- + # The Subconscious OpenHuman's subconscious is what separates it from every other AI agent. Most AI systems follow the conscious mind model: they wait for instructions and execute them. They cannot think on their own. OpenHuman can. diff --git a/gitbooks/use-cases/community-managers-and-moderators.md b/gitbooks/use-cases/community-managers-and-moderators.md new file mode 100644 index 000000000..58814f975 --- /dev/null +++ b/gitbooks/use-cases/community-managers-and-moderators.md @@ -0,0 +1,33 @@ +--- +icon: '5' +--- + +# Community Managers & Moderators + +Your community has scaled to hundreds or thousands of members. You cannot read every message. Sentiment shifts, conflicts, and manipulation happen faster than you can monitor. + +#### The problem + +Moderation at scale is reactive. You catch problems after they blow up. No systematic way to track community health over time. Burnout is constant. + +#### How OpenHuman helps + +**Aggregated community health.** Surfaces patterns across your entire community: topic trends, participation distribution, engagement, and sentiment. + +**Early warning signals.** Coordinated negative sentiment, scam patterns, and unusual account activity surfaced before escalation. + +**Contribution recognition.** Identifies constructive members based on behavioral evidence. + +**Moderation context.** Full history on any member: past behavior, interaction patterns, whether an incident is isolated or part of a pattern. + +#### Example prompts + +* "How is community health this week vs. last week?" +* "Flag any unusual patterns in the last 48 hours" +* "Who are the top contributors this month?" +* "Give me the full context on this member's activity" +* "Are there any emerging conflicts I should be aware of?" + +#### When this use case is strongest + +Best when managing 100+ active members across multiple channels. If you can read every message, the setup overhead may not be worth it. diff --git a/gitbooks/use-cases/dao-and-web3-communities.md b/gitbooks/use-cases/dao-and-web3-communities.md new file mode 100644 index 000000000..4c7cb152c --- /dev/null +++ b/gitbooks/use-cases/dao-and-web3-communities.md @@ -0,0 +1,32 @@ +--- +icon: '4' +--- + +# DAO and Web3 Communities + +Your DAO makes governance decisions through messaging. Proposals are discussed in Telegram, debated in Discord, and eventually formalized in votes. As the community grows, institutional memory degrades. + +#### The problem + +Chat-based governance creates accountability gaps. Decisions are scattered across threads. Contributors come and go. The same debates repeat because nobody can find the original discussion. + +#### How OpenHuman helps + +**Decision preservation.** Decisions are extracted as structured records with attribution, rationale, and source context. Surfaceable months later. + +**Contribution tracking.** Participation patterns tracked across conversations: who is active, constructive, and following through. + +**Governance continuity.** New proposals automatically surface the original context of referenced past decisions. + +**Community health signals.** Aggregated participation, topic trends, and early warning signals for governance fatigue. + +#### Example prompts + +* "What governance decisions were made in the last quarter?" +* "Show me the discussion that led to the treasury allocation vote in March" +* "Who were the most active contributors this month?" +* "Have we discussed this topic before? Show me the context." + +#### When this use case is strongest + +Best when governance happens primarily in messaging with 50+ active participants. Smaller DAOs can usually keep context in their heads. diff --git a/gitbooks/use-cases/distributed-teams.md b/gitbooks/use-cases/distributed-teams.md new file mode 100644 index 000000000..46fe32936 --- /dev/null +++ b/gitbooks/use-cases/distributed-teams.md @@ -0,0 +1,32 @@ +--- +icon: '6' +--- + +# Distributed Teams + +Your team makes decisions in chat and struggles to execute. Commitments dissolve into scroll history. Contradictions between separate conversations go unnoticed until standups. + +#### The problem + +Chat is fast but ephemeral. Action items live in threads nobody revisits. Accountability depends on individual memory, which fails at scale. + +#### How OpenHuman helps + +**Decision and action extraction.** Commitments become structured records with attribution and context. + +**Contradiction detection.** The subconscious flags conflicting commitments across separate conversations before they surface as surprises. + +**Structured exports.** Decisions and action items exportable without manual transcription. + +**Cross-channel context.** Unified answers spanning Slack, Telegram, and screen activity. + +#### Example prompts + +* "What did we commit to this week across all channels?" +* "Are there any contradictions between what different team members said?" +* "Extract all action items from #product since Monday" +* "What decisions were made yesterday that need documenting?" + +#### When this use case is strongest + +Best when coordination happens through messaging across multiple channels with 5+ people. If your team uses a rigid project management tool for all commitments, OpenHuman adds less value. diff --git a/gitbooks/use-cases/individual-power-users.md b/gitbooks/use-cases/individual-power-users.md new file mode 100644 index 000000000..714b6a3b3 --- /dev/null +++ b/gitbooks/use-cases/individual-power-users.md @@ -0,0 +1,40 @@ +--- +icon: '2' +--- + +# Individual Power Users + +You live in messaging. Dozens of active Telegram groups, Slack channels, DMs, and threads. Every morning starts with a wall of unread messages. The important stuff is buried under memes, noise, and repetitive discussions. + +#### The problem + +High-volume messaging creates information overload that gets worse as you join more groups. Reading everything is impossible. Skimming means you miss decisions, commitments, and context that matter. + +#### How OpenHuman helps + +**Compression turns noise into signal.** Neocortex strips 80%+ redundant messages from group chats through semantic deduplication. What remains: decisions, action items, risks, sentiment shifts, and things that need your attention. + +**Briefings replace scrolling.** Ask one question, get a compressed summary. 247 messages become a 30-second read. + +**Nothing falls through the cracks.** The subconscious catches buried context: approaching deadlines, unanswered questions directed at you, sentiment shifts in communities you moderate. + +#### Example prompts + +* "What do I need to know from the last 12 hours?" +* "Are there any messages waiting for my response?" +* "Summarize the key decisions from #engineering this week" +* "What are the most heated discussions across my groups right now?" +* "Extract all action items assigned to me from the last 3 days" + +#### Features that matter most here + +| Feature | Why it matters | +| ------------------------- | ------------------------------------------------------------ | +| Neocortex compression | Strips noise, surfaces signal from thousands of messages | +| Subconscious intelligence | Catches buried context and approaching deadlines proactively | +| Telegram integration | Full access to chats, search, history, and message actions | +| Signal extraction | Decisions, action items, and risks identified automatically | + +#### When this use case is strongest + +Power Users get the most value when active in 10+ conversations. If you have 2-3 quiet channels, the manual approach works fine. diff --git a/gitbooks/use-cases/knowledge-workers.md b/gitbooks/use-cases/knowledge-workers.md new file mode 100644 index 000000000..b3f30a584 --- /dev/null +++ b/gitbooks/use-cases/knowledge-workers.md @@ -0,0 +1,60 @@ +--- +icon: '1' +--- + +# Knowledge Workers + +## + +*** + +### Knowledge Workers + +You work across 8+ applications every day. Email, Slack, spreadsheets, documents, browsers, design tools, project boards. Each app switch costs you context. By the end of the day, the thread connecting your morning research to your afternoon decisions has frayed. You spend as much time reconstructing context as you do acting on it. + +OpenHuman turns that fragmented workflow into cumulative awareness. + +#### The problem + +Context evaporates with every app switch. You review a spreadsheet at 10am, discuss it in Slack at 11am, and by 2pm when you need to write a summary email, you have to reopen the spreadsheet, re-read the Slack thread, and piece it all back together. + +Your brain does this all day. It is exhausting, and things fall through the cracks. + +#### How OpenHuman helps + +**Screen Intelligence tracks your work across applications.** Every app you use, every document you view, every conversation on screen is captured and summarized locally. You build a continuous record of your workday without logging anything manually. Capture frequency is configurable (default: 1 frame per second) and sensitive apps can be excluded. + +**Inline Autocomplete reflects your real context.** When you start typing that summary email at 2pm, the completions already know the Q3 numbers from the spreadsheet, the concerns raised in Slack, and the names of the people involved. Your typing becomes faster and more accurate because the AI has the same context you do. + +**Cross-source queries bridge your tools.** Instead of opening three apps to reconstruct a decision chain, you ask one question: "What happened with the Q3 projections today?" and get a unified answer spanning your screen activity, messages, and documents. + +#### Example prompts + +* "What was I working on before the standup?" +* "Summarize everything I saw related to Project Atlas today" +* "What was on that dashboard I was looking at this morning?" +* "What did the team discuss about the launch timeline while I was heads-down in the doc?" +* "Connect what Sarah said in Slack with the spreadsheet I was reviewing" + +#### Features that matter most here + +| Feature | Why it matters | +| -------------------- | ------------------------------------------------------------------- | +| Screen Intelligence | Builds continuous work context across all your apps | +| Inline Autocomplete | Speeds up writing by reflecting your actual project context | +| Neocortex memory | Retains context across days and weeks, not just the current session | +| Cross-source queries | One question spans screen activity, messaging, and documents | + +#### A typical workflow + +**Morning:** You open your laptop. Screen Intelligence starts capturing as you check email, review a Jira board, scan Slack channels, and open a Google Doc. + +**Midday:** You ask "What are the open threads I need to respond to?" OpenHuman returns: two Slack messages waiting for reply, an email you read but did not respond to, and a comment on the Google Doc you were editing. + +**Afternoon:** You start drafting a project update. Inline Autocomplete suggests phrasing that references this morning's spreadsheet data and yesterday's Slack decisions. You finish in half the time. + +**End of day:** You ask "What did I commit to today?" and get a clean list with source attribution. + +#### When this use case is strongest + +Knowledge Workers get the most value when they work across many applications daily and frequently need to reference earlier work. If your workflow is single-app, Screen Intelligence adds less value. diff --git a/gitbooks/use-cases/overview.md b/gitbooks/use-cases/overview.md index 5d39ea644..6db856374 100644 --- a/gitbooks/use-cases/overview.md +++ b/gitbooks/use-cases/overview.md @@ -1,3 +1,7 @@ +--- +icon: grid-4 +--- + # Overview OpenHuman adapts to different coordination and information environments across communication platforms. While the underlying system remains the same, the way value is realized depends on how your tools and conversations are used in each context. This section outlines common use cases to help you understand where OpenHuman fits naturally and where it may offer limited benefit. diff --git a/gitbooks/use-cases/service-providers-and-contributors.md b/gitbooks/use-cases/service-providers-and-contributors.md new file mode 100644 index 000000000..bcb98a42a --- /dev/null +++ b/gitbooks/use-cases/service-providers-and-contributors.md @@ -0,0 +1,30 @@ +--- +icon: '7' +--- + +# Service Providers & Contributors + +You work across multiple communities and projects as a freelancer, consultant, or contributor. Your track record is your most valuable asset, but it is fragmented across platforms. + +#### The problem + +Reputation in decentralized ecosystems is hard to build. Your best work is buried in private channels and resolved issues. Proving reliability to new clients means assembling scattered evidence manually. + +#### How OpenHuman helps + +**Behavioral contribution tracking.** Participation patterns tracked over time: delivery consistency, engagement quality, expertise areas, and follow-through. + +**Portable credibility.** Contribution history moves with you across communities. + +**Scope documentation.** Decisions, deliverables, and outcomes extracted from project conversations automatically. + +#### Example prompts + +* "Summarize my contributions across all communities this month" +* "What projects have I delivered on in the last quarter?" +* "What commitments have I made that are still open?" +* "Draft a project summary from my conversations with the Meridian team" + +#### When this use case is strongest + +Best when working across 3+ communities or projects simultaneously. If you work with one team on one project, a personal log may suffice. diff --git a/gitbooks/use-cases/traders-and-analysts.md b/gitbooks/use-cases/traders-and-analysts.md new file mode 100644 index 000000000..06d5aaef6 --- /dev/null +++ b/gitbooks/use-cases/traders-and-analysts.md @@ -0,0 +1,43 @@ +--- +icon: '3' +--- + +# Traders and Analysts + +You monitor multiple Telegram groups, Discord channels, Twitter feeds, and on-chain activity for alpha. The noise-to-signal ratio is brutal. + +#### The problem + +Trading groups move fast. By the time you verify a claim, the opportunity has passed. Acting on unverified claims is how you lose money. Scams and coordinated manipulation are constant threats. + +#### How OpenHuman helps + +**Claim tracking with resolution.** Predictions are extracted as structured claims with resolution dates. Over time, you build track records for every participant. + +**Risk intelligence.** Behavioral pattern recognition flags coordinated message timing, impersonation, and repeated failure patterns across groups. + +**Historical context at query speed.** "What did this group say about SOL in January?" returns compressed history from months of indexed data. + +**On-chain + off-chain fusion.** When wallet data is connected, Neocortex correlates on-chain activity with off-chain discussions. + +#### Example prompts + +* "What claims were made about ETH this week and who made them?" +* "Show me the track record of @cryptoalpha in the trading group" +* "Flag any suspicious patterns across my groups in the last 24 hours" +* "Summarize the alpha from my top 5 groups today" +* "Who has been consistently accurate about L2 predictions?" + +#### Features that matter most here + +| Feature | Why it matters | +| ------------------------------ | --------------------------------------------------------------- | +| Claim tracking | Predictions become verifiable records | +| Risk intelligence | Flags scam patterns and coordinated abuse | +| Temporal context | Recalls historical discussions and correlates with price action | +| Cross-source entity resolution | Links on-chain events with off-chain conversations | +| Semantic deduplication | Strips repetitive alpha regurgitation | + +#### When this use case is strongest + +Best when monitoring 5+ information sources simultaneously. If you trade on technical analysis alone without social signals, this adds less value.