Everything you need to know about OsirisBrain, from installation to mastering your AI team.
OsirisBrain is a desktop AI assistant that runs locally on your computer. It connects to AI providers like OpenAI (ChatGPT), Anthropic (Claude), and Google (Gemini) to give you a team of 12 specialized professionals, all accessible from one app.
OsirisBrain supports 10+ AI providers including OpenAI, Anthropic, Google Gemini, Groq, xAI, Mistral, OpenRouter, DeepSeek, Together, NVIDIA, and Ollama. Setting it up takes less than a minute; see the Settings & Configuration section below.
Go to the download section and pick "Apple Silicon" (for M1/M2/M3/M4 Macs) or "Intel" (for older Macs). Not sure? Click the Apple logo → "About This Mac" , if it says "Apple M1" (or M2, M3, M4), you have Apple Silicon.
Double-click the downloaded file. A window appears with the OsirisBrain icon and your Applications folder. Drag the icon into Applications.
Open your Applications folder and double-click OsirisBrain. The first time, macOS may ask: "Are you sure you want to open it?" Click Open. This only happens once.
Go to the download section and click "Windows". The .exe installer will download.
Double-click the downloaded file. Follow the installation wizard. You can choose where to install it or just use the default location. Click "Install" and wait a few seconds.
OsirisBrain will appear on your Desktop and Start Menu. Double-click it to launch. Windows Defender may show a SmartScreen warning the first time. Click "More info" → "Run anyway".
Go to the download section and click "Linux". The .AppImage file will download.
Open a terminal and run:
chmod +x OsirisBrain-2026.2.17.AppImage
Double-click the AppImage file, or run it from the terminal:
./OsirisBrain-2026.2.17.AppImage
No installation needed. AppImages are portable and run directly.
Click the Apple icon in the top-left corner of your screen → "About This Mac". If it says Chip: Apple M1 (or M2, M3, M4), download the "Apple Silicon" version. If it says Processor: Intel, download the "Intel" version.
When you open OsirisBrain for the first time, here's what happens:
On first launch, OsirisBrain opens directly to the License panel in the settings sidebar. Paste your license key and click Activate. Once activated, the app switches to the Console.
Navigate to Providers in the settings sidebar. Add your API key and select a model. Use the engine toggle switch in the top-right corner to start the AI engine.
The professional lobby appears with all 12 specialists. Pick one to start chatting! We recommend starting with Osiris (the all-in-one assistant) if you're not sure.
OsirisBrain comes with 12 pre-built AI professionals. Each one is tuned for a specific skill set, with its own system prompt and personality. Here's your team:
Your all-in-one assistant. Does everything. The recommended starting point.
Code, deploy, automation. Best for programming questions and debugging.
Contracts, regulations, legal protection. Explains legal concepts in plain language.
Strategy, copy, growth hacking. Helps with campaigns and content strategy.
Health guidance and wellness. General health information (not a replacement for real medical advice).
Scheduling, emails, total organization. Helps you stay on top of tasks.
Explains, teaches, evaluates knowledge. Perfect for learning anything new.
Kids, homework, family activities. A guide for parenting challenges.
Finances, taxes, financial analysis. Helps with budgets and financial planning.
Creativity, visual design, branding. Advice on design choices and aesthetics.
Content, copywriting, storytelling. Helps craft text for any purpose.
Troubleshooting, setup, IT solutions. Fixes tech problems step by step.
When you select a professional, two things happen behind the scenes:
When you switch from one professional to another, OsirisBrain automatically resets the AI's memory. This prevents the previous professional's personality from "bleeding" into the new one. Each professional starts fresh.
Don't see the exact expert you need? Build your own! Custom professionals let you define the skills, personality, and knowledge of a completely new AI assistant.
In the professional lobby, click the "+ Create Professional" card at the end of the grid. The builder panel opens.
Name: What to call your professional (e.g., "Fitness Coach").
Description: A short tagline (e.g., "Workout plans and nutrition
advice").
Personality: How it should behave , friendly, strict, casual, formal,
etc.
Skills are specialized abilities. For example, a Fitness Coach might have skills called "Workout Programming" and "Nutrition Planning". Give each skill a title and detailed content describing what the AI should know.
Click Save. Your custom professional appears in the lobby alongside the built-in ones. Select it to start chatting!
Personality controls HOW the AI talks (tone, style, formality). Skills control WHAT the AI knows (expertise, specific knowledge). For best results, write clear, specific skills , the more detail you give, the better the AI performs.
The chat is where you interact with your chosen professional. It works like any messaging app, but with AI superpowers.
Your conversations are automatically saved. When you return to a professional you've talked to before, your previous messages will be loaded. Each professional has its own separate history.
OsirisBrain also maintains a persistent memory system that remembers key facts, preferences, and context across sessions. The agent automatically decides what to store based on conversation context , no configuration needed. Memory is stored locally on your machine using local embeddings for maximum privacy. The agent can search through past memories to inform future responses, making each interaction smarter over time.
When the AI replies, you'll see thumbs-up and thumbs-down buttons appear when you hover over the message. Click one to give feedback:
Your ratings directly influence how the agent responds in the future. The agent stores your feedback locally and uses it to adapt its behavior over time , learning whether you prefer more detailed or concise answers, code-first or explanation-first responses. Each professional learns independently based on your feedback.
The first time you use a professional, you'll see clickable suggestion chips , quick-start prompts tailored to that professional's expertise. Click one to instantly send it as your first message.
You can switch professionals at any time by going back to the lobby. When you switch, the AI's context resets automatically , the new professional won't remember what the previous one said. This is intentional: each professional starts with a clean slate.
Don't want to install anything? Use OsirisBrain directly in your browser at osirisbrain.com/chat.html.
The desktop app runs a local AI engine that connects directly to the AI provider for faster response times. The web chat works anywhere but may feel slightly slower. Both share the same license, professionals, and chat history.
OsirisBrain connects to the AI provider of your choice using your own API key. You only need to set this up once. Use the settings sidebar on the left to navigate between Console, Providers, Media, Budget, License, Agent, and Calibration. The engine toggle switch in the top-right corner starts and stops the AI engine.
OsirisBrain supports 25+ AI providers out of the box. Pick any provider below, sign up for an API key, and paste it into the app. Most offer free tiers or credits.
One API key gives you access to models from many providers.
Run models on your own hardware , no API key or cloud subscription needed.
In the settings sidebar on the left, click Providers to configure your AI connection.
Choose your AI provider from the dropdown. All providers listed above are available, plus any OpenAI-compatible custom endpoint.
Paste the API key you copied from your provider into the "API Key" field. The app encrypts and stores it locally on your device , it never leaves your computer.
Select your preferred model from the dropdown. Click Save and the engine restarts automatically with your new provider.
Select "Custom" as your provider if you want to use an OpenAI-compatible API not listed above. Enter the base URL and your API key. This works with any service that follows the OpenAI API format, including self-hosted inference servers like vLLM and text-generation-inference.
If you need to change the email associated with your license key, follow these steps:
Note: Your license key remains the same — only the associated email changes. The new email must not already be linked to another license. You must have an active device activation to change your email (this prevents unauthorized changes).
Your agent knows which plan you're on and how much time you have left today. This means it adapts its work strategy to be efficient within your daily budget:
This intelligence is built into the agent's system prompt and updates every time the engine starts. You don't need to configure anything — it works automatically.
The Protection panel is in the settings sidebar. It lets you set up Guardian Mode (for elderly users) or Parental Mode (for children). Once activated, the Protection Engine runs in the background and cannot be turned off without a PIN.
In the settings sidebar on the left, click Protection. You'll see two tabs at the top: Guardian and Parental. Pick the one you need.
Click the toggle switch next to the profile name. Guardian Mode is for protecting elderly users from scams and phishing. Parental Mode is for keeping children safe online.
Enter a 4-digit PIN and type your license email to confirm. This PIN is required to turn off protection later. Do not share it with the person being protected.
Click Activate Protection. The status badge at the top will change to show the active mode. The engine lock icon will appear, meaning the engine cannot be stopped without the PIN.
Once activated, OsirisBrain adds itself as a login item so it starts automatically when the computer turns on. The engine starts automatically too. If someone closes the app or restarts the computer, protection resumes on the next boot.
Guardian Mode protects elderly users from phishing, scams, and online fraud. After enabling the toggle, you can configure these settings:
Parental Mode keeps children safe by blocking harmful websites, monitoring browser activity, and enforcing screen time limits. After enabling the toggle, you can configure:
Under the Parental tab, you'll see category toggles. Turn each one on or off depending on what you want to block:
The engine uses community blocklists that are downloaded and merged automatically. These include Steven Black, OISD, URLhaus, and SomeoneWhoCares — covering hundreds of thousands of known harmful domains.
Below the category toggles, you'll see two lists: Blocked Domains and Allowed Domains.
example.com)
in the block field and click Add. This site will be blocked even if it's
not on any community blocklist.
In the Parental tab, look for the Screen Time section. Here you can set:
The Protection Engine writes blocked domains to the system's hosts file
(/etc/hosts on macOS/Linux, the equivalent on Windows). It also runs a
browser monitor that checks the active browser tab every few seconds. If a blocked site
is detected, the browser is redirected. A process watcher also runs to detect and close
restricted applications.
Click the Activity Log tab at the top of the Protection panel. This shows every action the engine has taken: blocked sites, flagged messages, screen time warnings, and more. Each entry includes a timestamp and description.
Click Protection in the settings sidebar. You'll see the active mode's status badge at the top.
Click the Deactivate Protection button. A confirmation dialog will appear asking for your PIN and email.
Type the 4-digit PIN you set during activation and your license email. Both must match exactly. Click Confirm to deactivate.
There is no PIN recovery. This is intentional — it prevents the protected person from bypassing the protection. If you forget your PIN, you will need to reinstall OsirisBrain and activate with a new PIN.
OsirisBrain includes a brain-inspired persistent memory system called Cortex. Your agent remembers facts, events, workflows, commitments, and preferences across sessions. No configuration required.
Cortex organizes memory into specialized regions, similar to the human brain:
| Region | Brain Area | What It Stores |
|---|---|---|
| Semantic | Temporal Lobe | Facts, identity, project details, technology stack, user rules |
| Episodic | Hippocampus | Conversations, decisions, events. Timestamped and searchable |
| Procedural | Cerebellum | Workflows, deployment steps, coding habits, learned patterns |
| Emotional | Amygdala | Importance tagging. Critical priorities surface first in recall |
| Commitments | Prefrontal Cortex | Active tasks, deadlines, responsibilities. Tracked until done |
| User Model | Social Cortex | Your preferences, communication style, and work habits |
| Failures | Error Memory | Past mistakes and bugs. Your agent avoids repeating them |
| Decisions | Basal Ganglia | Logged decisions with risk scores and outcomes |
Your agent has five dedicated tools for managing its memory:
| Tool | What It Does | Example |
|---|---|---|
| cortex_save | Stores a new memory in the right brain region | "Remember that our API uses port 3001" |
| cortex_recall | Searches memories by keyword across all regions | "What do you know about our deployment process?" |
| cortex_commit | Creates a new task or responsibility to track | "We need to fix the login bug by Friday" |
| cortex_complete | Marks a tracked commitment as done | "The login bug is fixed, mark it done" |
| cortex_briefing | Returns a full overview of active work and recent history | "What are we working on?" |
All memory data is stored locally on your machine in ~/.osirisbrain/workspace/cortex/ as
human-readable JSONL files. Nothing leaves your computer. You can inspect, edit, or delete any
memory file directly.
OsirisBrain comes with 100+ built-in tools that let your AI professionals take real actions on your behalf. Here's everything your agent can do; just ask in natural language.
Full control over your computer. Run commands, take screenshots, automate repetitive tasks.
| Tool | What it does | Example prompt |
|---|---|---|
| Shell | Runs terminal commands on your machine | "List all PDF files on my Desktop" |
| Desktop Control | Takes screenshots, moves mouse, types text | "Take a screenshot of my screen" |
| Clipboard | Reads and writes to your clipboard | "What's in my clipboard right now?" |
| Automation | Creates and runs AppleScript/JXA automations | "Set my Mac to dark mode" |
Send messages, manage contacts, and draft emails , all hands-free.
| Tool | What it does | Example prompt |
|---|---|---|
| Sends emails via SMTP (Gmail, Outlook, etc.) | "Send an email to John about the meeting tomorrow" | |
| Messaging | Sends iMessages and SMS, reads conversations | "Send Maria a message saying I'll be there at 3" |
| Contacts | Searches, creates, and manages macOS contacts | "Find the phone number for Carlos García" |
Work with any file on your system , read, write, search, convert, and organize.
| Tool | What it does | Example prompt |
|---|---|---|
| File Manager | Reads, writes, moves, copies, and lists files | "Create a new folder called Project X on my Desktop" |
| PDF Reader | Extracts text and metadata from PDF files | "Summarize the contract at ~/Documents/contract.pdf" |
| Document Editor | Creates and edits text documents, markdown files | "Write a meeting summary and save it as notes.md" |
Full internet access , your AI can search, browse, and call any API endpoint.
| Tool | What it does | Example prompt |
|---|---|---|
| Web Search | Searches the internet for current information | "What's the weather in Miami today?" |
| Web Fetch | Reads and extracts content from any webpage | "Read the article at this URL and summarize it" |
| API Client | Calls any REST or GraphQL API with custom headers | "Get Bitcoin price from CoinGecko API" |
| Browser | Controls a browser , clicks, types, navigates pages | "Open Google and search for flight deals to NYC" |
Manage your schedule, notes, and recurring tasks without leaving the chat.
| Tool | What it does | Example prompt |
|---|---|---|
| Calendar | Creates, lists, and manages calendar events | "Schedule a call with Ana for Friday at 2pm" |
| Notes | Creates, reads, and searches Apple Notes | "Create a note with my grocery list" |
| Cron Scheduler | Sets up recurring tasks and reminders | "Remind me to check email every morning at 9am" |
Find businesses, get directions, and explore places , all from your chat window.
| Tool | What it does | Example prompt |
|---|---|---|
| Maps Search | Finds businesses and places by name or category | "Find Italian restaurants near downtown Miami" |
| Directions | Gets directions and opens them in Apple Maps | "Give me directions from here to the airport" |
| Nearby | Searches for nearby places by category | "Find the nearest pharmacy" |
Run code, generate images, and build a personal knowledge base that grows with every conversation.
| Tool | What it does | Example prompt |
|---|---|---|
| Code Interpreter | Runs Python, JavaScript, and other code locally | "Calculate compound interest on $10,000 at 5%" |
| Image Generator | Creates images using AI models (DALL-E, etc.) | "Generate a logo for a coffee shop called Bean Co" |
| Knowledge Base | Stores and retrieves information across sessions | "Remember that my server IP is 192.168.1.50" |
| Canvas | Creates visual diagrams and rich content | "Draw a flowchart of my login process" |
Connect your AI to smart home devices, databases, sub-agents, and more.
| Tool | What it does | Example prompt |
|---|---|---|
| Smart Home | Controls HomeKit / Home Assistant devices | "Turn off all the lights in the living room" |
| Database | Queries SQLite and PostgreSQL databases | "Show me all users who signed up last month" |
| Sub-Agents | Spawns specialized sub-agents for complex tasks | "Research competitors and write a report" |
| Text-to-Speech | Reads text aloud using system voice synthesis | "Read this paragraph out loud" |
| Speech-to-Text | Transcribes voice input into text using OpenAI Whisper, Google Cloud, or local system speech recognition | "Listen to what I'm about to say" |
| Plugin Manager | Installs, removes, and manages plugins from the marketplace via natural language | "Install the web scraper plugin" |
| Memory | Stores and retrieves facts, preferences, and context across conversations using local embeddings | "Remember that I prefer dark mode" |
20 family-focused tools for managing your household, children, pets, health, and more — all from chat.
| Tool | What it does | Example prompt |
|---|---|---|
| Family Calendar | Shared family scheduling with color-coded members and conflict detection | "Add soccer practice for Emma on Tuesdays at 4pm" |
| Chore Manager | Gamified household tasks with points, streaks, and family leaderboard | "Assign dish washing to Lucas and give him 10 points" |
| Meal Planner | Weekly meal plans, grocery lists, allergy tracking, and recipe suggestions | "Plan this week's meals — Maria is allergic to gluten" |
| Family Budget | Expense tracking, allowances, savings goals, and spending reports | "How much have we spent on groceries this month?" |
| Kids Education | Academic tracking, homework management, grade monitoring | "What grades does Emma have in Math this semester?" |
| Emergency Contacts | Medical info, blood types, doctor numbers, emergency cards | "What is Lucas's blood type and pediatrician phone?" |
| Pet Care | Vet visits, feeding schedules, medication reminders, weight tracking | "When is Buddy's next vet appointment?" |
| Home Inventory | Item tracking, warranties, insurance reports, purchase history | "When does the TV warranty expire?" |
| Family Stories | Heritage preservation, traditions, family tree, and storytelling | "Record grandma's recipe for arroz con pollo" |
| School Hub | Parent-school communication, report cards, school events | "When is the next parent-teacher conference?" |
| Family Health | Wellness tracking, medications, immunization records, doctor visits | "Is Emma up to date on her vaccinations?" |
| Home Maintenance | Repair scheduling, seasonal checklists, contractor contacts | "When was the last AC filter change?" |
| Family Travel | Trip planning, packing lists, itineraries, passport expiration tracking | "Create a packing list for our trip to Orlando" |
| Child Milestones | Baby/child development tracking, growth charts, milestone logging | "Log that baby Sofia started walking today" |
| Coding Tutor | Interactive coding lessons for kids with real code execution | "Teach Lucas how to make a website with HTML" |
| Elder Companion | Elderly assistant with medication reminders, companionship, and safety | "Remind grandpa to take his blood pressure medicine at 8am" |
| Child Safety | Screen time limits, safe browsing rules, stranger danger alerts | "Set a 2-hour screen time limit for Emma" |
| Daily Routines | Morning/evening routines, streak tracking, habit building | "Create a morning routine for the kids" |
| Family Games | Trivia, word games, scavenger hunts, game night planning | "Start a family trivia night about geography" |
| Smart Security | System integrity checks, self-repair, lockdown mode | "Run a security audit on the system" |
A complete AI school with 10 specialized teachers, K–University curriculum, and professional-grade academic records. Uses the Socratic method (question-driven learning) and Pareto 80/20 principle (focus on the critical 20% that delivers 80% of results).
| Action | What it does | Example prompt |
|---|---|---|
| Lesson | Adaptive lesson with Bloom's Taxonomy progression and Socratic questioning | "Teach me about photosynthesis at a high school level" |
| Quiz | Auto-generated quiz with multiple formats (MCQ, open, true/false) | "Quiz me on algebra — 10 questions" |
| Flashcards | Leitner 5-box spaced repetition system for long-term retention | "Create flashcards for Spanish vocabulary" |
| Explain | Deep explanation with analogies and real-world examples | "Explain quantum computing like I'm 12" |
| Study Plan | Generates a personalized multi-day study plan for any subject | "Create a 2-week study plan for SAT Math prep" |
| Homework Help | Step-by-step guidance without giving away the answer | "Help me solve this calculus problem step by step" |
| Progress | Shows mastery levels, strengths, weaknesses, and advancement | "Show my progress in Science" |
| Transcript | Official GPA transcript with credit hours, Dean's List, and Honors | "Show my official academic transcript" |
| Session Log | Every learning session logged with date, time, teacher, duration, bloom level | "Show my learning sessions from this week" |
| Attendance | 30-day attendance calendar with daily hours, subjects, and streak tracking | "Show my attendance record" |
| Academic Record | Total hours, lessons, quizzes by subject with weekly/monthly trends | "Show my complete academic record" |
Prof. Logic (Mathematics) · Prof. Darwin (Science) · Prof. Shelley (Language Arts) · Prof. Herodotus (History) · Prof. Turing (Computer Science) · Prof. Picasso (Arts) · Prof. Aristotle (Philosophy) · Prof. Curie (Chemistry/Physics) · Coach Vitalis (Health/PE) · Sensei Lingua (Foreign Languages)
You don't need to know which tool to use — just describe what you want in plain English. OsirisBrain automatically selects the right tool(s) and chains them together. For example: "Find Maria's email in my contacts and send her a reminder about tomorrow's meeting" will use Contacts → Email in sequence.
Speed up your workflow with these shortcuts:
| Action | macOS | Windows |
|---|---|---|
| Send message | Enter | Enter |
| New line in message | Shift + Enter | Shift + Enter |
| Focus chat input | ⌘ + L | Ctrl + L |
| Go back to lobby | Esc | Esc |
Skills are reusable prompt fragments that define what your custom professionals can do. Each skill has a title and a content block — the actual prompt text that shapes the AI's behavior.
When you chat with a professional that has skills attached, each skill's content is automatically injected into the AI's context. This shapes the AI's responses without you having to repeat instructions.
Each skill can be edited inline — click the title or content to modify it. Changes auto-save when you click outside the field. You can also remove individual skills with the trash icon.
Browse the for community-created professional bundles and skills. Each listing includes a description, rating, and download count. You can install skills directly into your custom professionals.
Control your AI spending with OsirisBrain's built-in budget manager. Set a monthly limit and track usage in real-time.
OsirisBrain supports connecting multiple AI providers simultaneously. You can assign different providers to different professionals for optimal results.
Your default provider is used for all professionals unless overridden. To change it, click the star icon next to any provider in the list. The system uses only the provider you choose — there are no hardcoded defaults.
Assign specific providers to individual professionals. For example, use Anthropic Claude for your coding professional and OpenAI for creative writing. Each professional can have its own provider + model combination.
The OsirisBrain Marketplace is a community hub for sharing and discovering AI skills, professional bundles, and workflow templates.
Rate any skill from 1 to 5 stars and leave a review. If you find inappropriate content, use the flag button to report it.
The OsirisBrain Community is where users share setups, ask questions, report bugs, and help each other get the most from their agents.
Click any post to see its full content and replies. You can reply to any post, upvote helpful responses, and mark answers as accepted for resolved questions.
Migrate your entire workspace between machines or back up your agent's knowledge.
.zip archive.zip archive