Trilium AI is a WordPress plugin ecosystem built specifically for Trilium Notes users. It bridges the gap between your personal knowledge management system and cutting-edge AI.
Through two tightly integrated core plugins — Trilium AI Chat and Trilium AI Agent — you can chat with AI, search your notes, have AI read and analyze your knowledge base, and even let AI agents autonomously carry out complex multi-step workflows, all from within your WordPress site. Additional Trilium AI sub-plugins for further feature expansion will be covered in upcoming articles.
Picture this: you type a command into a chat window on your WordPress site, and the AI automatically sifts through years of notes you’ve accumulated in Trilium Notes, finds the relevant content, grasps the context, and delivers an in-depth analysis — all without you having to copy and paste a single word.
Trilium AI isn’t just another AI chat tool. It’s a central hub that connects knowledge with intelligence, breathing new life into the content buried in your note library.
🏗️ The Big Picture: Two Plugins, One Ecosystem
Trilium AI consists of two core components that can run independently or work together to unlock even greater capabilities:
┌─────────────────────────────────────────────────────────┐
│ WordPress Site │
│ │
│ ┌───────────────────┐ ┌────────────────────┐ │
│ │ Trilium AI Chat │ ◄──► │ Trilium AI Agent │ │
│ │ │ │ │ │
│ │ 💬 Chat with AI │ │ 🤖 Let AI act │ │
│ │ 📝 Rich rendering│ │ 📂 Operate your KB│ │
│ │ 🔌 Multiple AIs │ │ 🔄 Automate flows │ │
│ │ ⚡ Live streaming │ │ 🧩 Multi-agent │ │
│ │ 🌐 Web search │ │ 🧠 Long-term memory│ │
│ │ 🧱 Workflow blocks│ │ │ │
│ └───────────────────┘ └────────────────────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌────────────────────────────────────────────────┐ │
│ │ Google Gemini · OpenAI/POE · Ollama (Local) │ │
│ │ OpenCode Agent · OpenClaw Agent Platform │ │
│ │ Trilium Notes KB · FileBrowser File Manager │ │
│ └────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘Trilium AI Chat is the foundation layer — install it on its own and your WordPress site gets a full-featured AI chat experience with workflow orchestration. Trilium AI Agent is the enhancement layer, stacking agent capabilities, deep knowledge base operations, OpenClaw agents, and persistent memory on top of Chat. Start with Chat, and upgrade to Agent anytime to unlock the full power.
💬 Trilium AI Chat — Your AI Command Center
Trilium AI Chat (currently at v5.1.1) embeds a feature-rich AI chat window into your WordPress site. It’s far more than a simple Q&A interface — it’s a full-fledged conversation hub with multi-model switching, multi-turn context memory, web search, and knowledge base integration.
🔌 One Window, Every Major AI at Your Fingertips
No more jumping between different AI platforms. In the Trilium AI Chat window, a simple dropdown lets you switch freely between multiple AI models:
☁️ Google Gemini Family — From the high-performance Gemini 2.5 Pro to the lightweight and fast Gemini 2.0 Flash, covering everything from deep analysis to everyday Q&A. Web search support means AI responses draw not only on training data but also on real-time information from the internet.
☁️ OpenAI / POE Compatible APIs — GPT-4o, Claude 3.7 Thinking, Grok 3 DeepSearch, Deepseek V3, and more of today’s most popular models, all accessible. Through the POE API, you also get access to Claude Sonnet 4.5, Claude Opus 4.5, Gemini 3.0 Pro, GPT-5.1, and other reasoning-enhanced models, with fine-grained control over “thinking budget” and “reasoning intensity” — when you hit a tough problem, give the AI more “thinking time” for a better answer.
🏠 Ollama Local Models — If data privacy is non-negotiable, connect fully local open-source models through Ollama. All data and inference stay on your own server — not a single byte leaves your network.
This flexible multi-model architecture means you can make the best choice for every conversation: use Claude Opus when you need top-tier reasoning, switch to Gemini Flash for speed, and fall back to local Ollama for sensitive information. Your model preference is remembered automatically — next time you open the window, it defaults to your last pick.
✨ Real-Time Streaming Responses
Nobody likes staring at a blank screen waiting for an AI to finish generating its answer. Trilium AI Chat uses streaming output — the AI’s response appears word by word, like a typewriter, right before your eyes. You can read and think as it goes, without wasting a second. A blinking cursor and dynamic loading indicator keep you informed that the AI is hard at work.
📝 Beautiful Content Rendering
AI responses shouldn’t be a wall of dense text. Trilium AI Chat has a full-featured Markdown rendering engine built in, automatically formatting every AI response into clean, readable rich text:
- 📊 Tables render as neat grids with horizontal scrolling on narrow screens
- 💻 Code blocks come with syntax highlighting and a one-click copy button — paste straight into your editor
- 📋 Lists, blockquotes, and headings are clearly structured, making even long responses easy to scan
- ✅ Task lists with checkboxes — perfect for AI-generated to-do lists
🎨 A Polished, Practical Interface
Trilium AI Chat’s interface balances aesthetics with usability, delivering a smooth experience on both desktop and mobile:
- 📱 Adaptive layout — The floating chat window sits elegantly in the bottom-right corner on desktop and automatically adjusts to fit the screen on mobile. One-click fullscreen mode is also available
- 🌓 Automatic dark mode — Follows your system theme preference, switching between light and dark automatically with no manual setup
- ⌨️ Efficient input — The text box expands as you type;
Entersends,Shift + Enteradds a new line; type/to bring up the command menu - 🔔 Elegant notifications — Action feedback slides in from the bottom as toast messages, never interrupting your flow
🎛️ You’re in Control of the AI’s Behavior
In the chat window’s options bar, you have full control over how the AI behaves:
- 💾 Save to Trilium — Toggle it on, and completed conversations are automatically archived to your Trilium Notes knowledge base. Never lose a valuable AI conversation again
- 🔗 Multi-turn context memory — Enable context mode and the AI remembers your recent conversation history (up to 20 messages), making your conversations truly continuous instead of starting from scratch every time
- 🌐 Web search — Let the AI consult the latest internet information before answering — especially useful for time-sensitive questions
- 🎚️ Reasoning depth control — For reasoning-enhanced models, a slider lets you fine-tune how deeply the AI thinks. Dial it down for quick answers to simple questions; crank it up for complex problems that demand deep reasoning
📂 Drive Everything with Commands
Beyond regular conversation, slash commands give you quick access to specific operations:
| Command | What You Can Do with It |
|---|---|
/trilium send | Save the current conversation to Trilium Notes in one click |
/trilium help | View all available commands |
/trilium custom_prompt_save | Save frequently used prompt templates for one-click reuse |
/trilium custom_prompt_delete | Manage your existing prompt templates |
The command system is open by design — install Trilium AI Agent and the command menu automatically expands with advanced commands for searching notes, browsing directories, exporting analyses, and more. Any WordPress plugin can register its own commands through the standard Hook mechanism, creating a continuously growing AI capability ecosystem.
🧱 Gutenberg Block — The Real Reason to Build AI into WordPress
If the chat window is a general-purpose AI conversation entry point, then the Gutenberg Block is the key move that deeply embeds AI capabilities into the WordPress content system — and the core reason we chose to build our AI platform within the WordPress ecosystem.
💡 Why WordPress?
There’s no shortage of AI chat tools out there. But Trilium AI’s Gutenberg Block does something fundamentally different: it lets you design and deploy AI workflows the WordPress way.
WordPress’s Gutenberg editor is inherently a visual content orchestration system — you drag and drop blocks to build pages. Trilium AI’s Gutenberg Block lets you orchestrate AI workflows in exactly the same way: place a pre-configured AI interaction point anywhere on a page, in any combination, for any scenario. No code required, no technical knowledge needed — if you can use the WordPress editor, you can design professional AI workflows.
That’s why integrating AI into WordPress matters: not to slap a chat window onto a webpage, but to gain a no-code AI workflow orchestration platform.
🔧 What Can It Actually Do?
Each Gutenberg Block instance is an independent, customizable AI workflow node. You can configure it with:
- 📌 Custom title — Give the AI interaction a context-specific name, like “Smart Contract Review,” “One-Click Summary Generator,” or “Multilingual Translation Assistant”
- 📝 Prompt template — Pre-define the specific task for the AI, such as “Analyze the key arguments in the following text and summarize the main points”
- 🔘 Button label — Customize the call to action, e.g., “Start Analysis,” “Translate Now,” “Generate Report”
- 💾 Save toggle — Independently control whether this workflow’s output is automatically saved to Trilium Notes
✏️ The Variable System: The Soul of AI Workflows
The Gutenberg Block’s most powerful feature is its prompt variable system. Use {variable_name} placeholders in your prompt template, and the system automatically generates input fields on the frontend. Users fill in the blanks, the system assembles the complete prompt, and sends it to the AI.
A real-world example: You create a “Contract Risk Review” Block with the following prompt template:
Please review the following contract clauses from a legal advisor’s perspective: {contract_clauses}. Focus on: {review_priorities}. Output a risk-level assessment and revision recommendations.
The end user sees two input fields — “Contract Clauses” and “Review Priorities” — fills them in, clicks “Start Review,” and the AI performs the task in its pre-assigned professional role. Users never need to know how to write prompts — you’ve already designed the optimal AI interaction flow for them.
What’s more, each variable’s display label is fully customizable. {contract_text} can appear as “Paste the contract clauses here,” and {focus_areas} can show as “What aspects would you like to focus on?” — letting non-technical users engage with sophisticated AI workflows at zero learning cost.
🧩 Multiple Workflows on a Single Page
Since each Block instance is independently configured, you can place multiple different AI workflows on the same WordPress page without any interference:
┌─────────────────────────────────────────┐
│ 📄 Internal Knowledge Base Page │
│ │
│ ┌─────────────────────────────────┐ │
│ │ 🔍 Smart Document Search │ │
│ │ "Search the knowledge base for: │ │
│ │ {keywords}" │ │
│ │ [Search] │ │
│ └─────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────┐ │
│ │ 📝 Meeting Minutes Generator │ │
│ │ "Organize these meeting notes: │ │
│ │ {meeting_content}" │ │
│ │ "Output format: {format}" │ │
│ │ [Generate Minutes] │ │
│ └─────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────┐ │
│ │ 🌐 Multilingual Translation │ │
│ │ "Translate the following into │ │
│ │ {target_language}: {content}" │ │
│ │ [Translate Now] │ │
│ └─────────────────────────────────┘ │
└─────────────────────────────────────────┘These three Blocks share the same underlying AI infrastructure (model selection, streaming responses, Trilium storage) but each runs a completely different workflow — one for search, one for content generation, one for translation. Drag and drop them in the WordPress editor, and you’ve built a multi-function AI workbench.
🎯 From “General Chat” to “Scenario-Specific Workflows”
The floating chat window is a general-purpose AI entry point — users ask whatever they want. The Gutenberg Block channels AI capabilities into specific scenarios, using pre-defined prompt templates and the variable system to guide users through particular tasks. This delivers three key benefits:
🎓 Lower the barrier to entry Not everyone knows how to craft effective prompts. Blocks let you, as the “workflow designer,” set up the optimal prompts and interaction flow in advance. End users just fill in a few input fields and get professional-grade AI output.
🔒 Consistent output quality The quality of free-form AI responses depends heavily on how well the user asks. With carefully designed prompt templates, Blocks ensure that every AI interaction follows a pre-defined role, format, and focus — delivering stable, reliable output quality.
📐 Align with business processes Every organization and team has unique workflows. The Block’s customizability lets you embed AI capabilities precisely into specific steps of those processes — instead of making people adapt to the tool, the tool adapts to the people.
🔄 True Pipeline Orchestration
The value of Gutenberg Blocks goes beyond individual AI interactions. When combined with Trilium AI’s other capabilities, they form a complete AI workflow pipeline:
📝 Block collects input ──► 🤖 AI executes task ──► 💾 Results saved to Trilium
│
▼
📂 Agent retrieves related notes
│
▼
🔄 Workflow engine: deep analysis
│
▼
🧠 OpenClaw accumulates memoryA user submits a contract in a Block → AI completes the review and outputs an analysis report → the report is automatically saved to Trilium Notes → Agent can retrieve this analysis in future work → the OpenClaw agent remembers the context of this review and can reference it the next time a similar contract comes up.
This isn’t an isolated Q&A session — it’s a complete knowledge workflow pipeline from input to long-term retention, from one-off interaction to cumulative learning. And it all starts with a simple Gutenberg Block in the WordPress editor.
🤖 Trilium AI Agent — Let AI Work for You
Trilium AI Agent (currently at v3.0.0) is the advanced extension layer of the Trilium AI ecosystem. If Chat lets you talk to AI, then Agent lets AI take action on your behalf. With Agent installed, AI is no longer a passive question-answerer — it can proactively search your notes, analyze document contents, manage files, and even complete a chain of complex tasks following pre-defined workflows.
📂 Deep Access to Your Knowledge Base
Agent’s most exciting capability is giving AI genuine insight into your Trilium Notes knowledge base. Here’s what you can do directly from the chat window:
- 🔍 Search notes — Enter keywords and the AI finds relevant notes across your entire knowledge base, like having a personal assistant who knows every note you’ve ever written
- 🌳 Browse the directory tree — Navigate your note hierarchy directly in the chat window without opening the Trilium client, drilling down through your folder structure layer by layer
- 📄 Fetch note contents — Specify one or more notes and the AI instantly retrieves the full text for reading, summarization, or further analysis
- 📤 Export as files — Export notes from Trilium as well-formatted Markdown files, complete with title, date, and other metadata, stored in FileBrowser for later use
- 💬 Chat based on your notes — Send note contents directly to the AI and have it answer questions, generate content, or offer suggestions grounded in your own accumulated knowledge
This means your years of accumulated notes are no longer static text sitting idle in a database — they become live knowledge the AI can tap into at any time.
🔄 One Command, Automated End-to-End
Agent’s standout feature is its workflow engine. It automates the entire sequence of “search notes → export as files → hand off to AI for analysis”:
🔍 Search ──► 📤 Export ──► 🤖 AI Analysis ──► 📊 Output ResultsA real-world scenario: Say you’re a researcher with hundreds of notes on “machine learning” in Trilium Notes. You enter a single workflow command, and Agent automatically searches for relevant notes, organizes and exports them as Markdown files, sends them to the AI agent for comprehensive analysis, and returns a content summary or research insight. The entire process takes just one command.
🧩 Multi-Agent Management: Dedicated AI for Different Tasks
You can create multiple independent AI agent instances, each with its own configuration, connections, and memory space. It’s like having an AI team where each member has a different specialty:
- 🖥️ Code review agent — Connected to your code repository’s FileBrowser, focused on code analysis and technical Q&A
- 📖 Writing assistant agent — Connected to your documentation knowledge base, helping you draft articles and organize source material
- 🔬 Research analyst agent — Connected to your academic notes, helping you review literature and distill key insights
- 📋 Project management agent — Tracking project-related notes, monitoring progress, and generating weekly reports
All agents are a single click away in the chat window — switch between them like moving between different AI specialists. Any settings you haven’t customized automatically inherit from the global defaults, saving you from redundant configuration.
🧠 AI with Long-Term Memory
Ordinary AI chats “forget” everything the moment you close the window. But through OpenClaw’s persistent memory system, your AI agents accumulate knowledge across session boundaries:
- 📖 Long-term memory — The AI continuously remembers your preferences, work style, and frequently used terminology. The longer you use it, the better it knows you
- 📅 Daily journals — Highlights from each day’s interactions are automatically archived as daily logs, forming the AI’s own “work notes”
- 🔄 Knowledge sync — Sync Trilium note contents into the AI’s memory so it can recall them at any time
- 🧹 Memory management — View, edit, or clear the AI’s memory whenever you want — you always stay in control
This means your AI assistant is no longer a stranger who starts from scratch every time, but a partner who genuinely knows you and grows alongside you over time.
🎮 A Rich Set of Extended Commands
After installing Agent, the chat window’s command menu automatically expands with powerful new capabilities:
| Command | What You Can Do with It |
|---|---|
/trilium agent chat <message> | Send a message to the AI agent for an in-depth conversation |
/trilium agent search <keywords> | Search your Trilium note library |
/trilium agent browse | Browse the Trilium note directory structure |
/trilium agent export <note_ID> | Export a specific note as a Markdown file |
/trilium agent new | Start a fresh agent work session |
/trilium agent sessions | View and manage all historical sessions |
/trilium agent switch <session_ID> | Switch freely between different sessions |
/trilium agent status | Check the current agent service status |
/trilium send | Quickly save content to Trilium Notes |
🐾 OpenClaw — The Agent Soul Behind Trilium AI
Within the Trilium AI ecosystem, OpenClaw plays a pivotal role. It’s not just another AI service provider — it’s a local AI agent runtime platform, and it’s precisely what gives Trilium AI the power to transcend the boundaries of a mere “chat tool.”
🌟 What Is OpenClaw?
OpenClaw is an emerging open-source AI agent gateway designed for running persistent, stateful AI agents in local environments. Unlike traditional “ask-and-answer” AI APIs, agents in OpenClaw are always running — they have their own identity, memory, and context, maintaining coherent cognitive state across multiple sessions.
Think of OpenClaw as a “management platform for AI agents”:
- 🏠 Locally deployed — Runs on your own device or server, with data entirely under your control
- 🧠 Stateful agents — Each agent has its own persistent memory that doesn’t vanish when you close the browser
- 🔐 Secure connections — Authentication via device identity pairing ensures only authorized devices can access your AI agents
- ⚡ Real-time communication — Messages arrive instantly, keeping conversations smooth and lag-free
- 🔗 Open ecosystem — Designed as a gateway layer that can plug into various AI models and services, with no vendor lock-in
🔗 OpenClaw × Trilium AI: 1 + 1 > 2
Trilium AI’s deep integration with OpenClaw creates a powerful synergy:
🧠 Remembers everything about you Ordinary AI chats start from zero every time. But through OpenClaw, your AI agents have persistent long-term memory and daily interaction logs. Tell it today that “I’m working on a sustainability research project,” and a week later it still remembers — offering more precise advice grounded in that context.
📂 Direct access to your knowledge base OpenClaw agents don’t just chat — they can operate on your Trilium Notes directly through Trilium AI’s workflow engine: searching notes, reading content, exporting files, syncing knowledge. Your note library becomes the AI agent’s “knowledge backend.”
🧩 Multi-agent collaboration Run multiple agents on OpenClaw, each with a different role and memory. Through Trilium AI’s multi-agent management system, switch between them in the chat window with a single click — like moving between different AI specialists.
🏠 Fully private deployment The combination of OpenClaw + Ollama local models + local FileBrowser means you can build an AI knowledge assistant system that depends on absolutely no cloud services. All data, all inference, all memory stays on your own infrastructure — total data sovereignty.
🛡️ Security & Privacy
Trilium AI provides multiple layers of protection:
- 🔐 Authentication — All operations require user login verification; unauthorized users cannot access any AI functionality
- 🔑 Encrypted communication — Streaming response connections use independent encrypted tokens that auto-expire and rotate every hour
- 🛡️ Device pairing — OpenClaw connections use encrypted device identity authentication, ensuring only your trusted devices can access AI agents
- 🏠 Local-first — With the Ollama + local OpenCode/OpenClaw combination, sensitive data can stay entirely on your server
- 🔒 Open-source transparency — Licensed under GPL v2+, the code is fully open — you can audit every line of data-handling logic
📊 Feature Comparison at a Glance
| Capability | Trilium AI Chat | + Trilium AI Agent |
|---|---|---|
| 💬 AI Conversations | ✅ Multi-model chat | ✅ + Agent conversations |
| 🔌 AI Models | Google · OpenAI/POE · Ollama | + OpenCode · OpenClaw |
| 📝 Content Rendering | ✅ Full Markdown rendering | ✅ Inherited |
| ⚡ Real-time Streaming | ✅ | ✅ |
| 🌐 Web Search | ✅ | ✅ |
| 💾 Save Conversations to Trilium | ✅ | ✅ |
| 🔗 Multi-turn Context Memory | ✅ Up to 20 messages | ✅ |
| 🧱 Gutenberg Workflow Orchestration | ✅ Variable templates + multi-instance | ✅ |
| 📂 Knowledge Base Search & Browse | ❌ | ✅ |
| 📤 Note Export & Analysis | ❌ | ✅ |
| 🔄 Automated Workflows | ❌ | ✅ |
| 🧩 Multi-Agent Management | ❌ | ✅ |
| 🧠 AI Long-Term Memory | ❌ | ✅ (via OpenClaw) |
| 🐾 OpenClaw Agents | ❌ | ✅ |
| 🎚️ Reasoning Depth Control | ✅ | ✅ |
| 📂 Custom Prompt Templates | ✅ | ✅ |
| 🌓 Dark Mode | ✅ Auto-adaptive | ✅ Auto-adaptive |
| 📱 Mobile-Friendly | ✅ | ✅ |
🚀 Getting Started
Just a few simple steps to bring AI capabilities to your WordPress site:
📋 Prerequisites
- ✅ WordPress 5.8 or later
- ✅ PHP 7.4 or later
- ✅ A running Trilium Notes instance (with ETAPI enabled)
- ✅ At least one AI service API key (Google Gemini / OpenAI), or a local Ollama installation
If you want to use the full capabilities of the Agent plugin:
- ✅ FileBrowser file management service (for note export and AI memory storage)
- ✅ OpenClaw Gateway or OpenCode service (for AI agent connectivity)
📥 Three-Step Installation
1️⃣ Install Trilium AI Chat — Upload and activate the plugin, then configure your AI service API keys and Trilium connection details in the admin dashboard. Once done, the AI chat window appears on your site’s frontend immediately — ready to use out of the box.
2️⃣ Install Trilium AI Agent (recommended) — Upload and activate, and it automatically detects the Chat plugin and injects its extended capabilities. Configure your OpenCode, OpenClaw, and FileBrowser connection details under “Trilium AI → Agent Settings.”
3️⃣ Start using it — Select your preferred AI model in the chat window, type /trilium help to see all available commands, or just start chatting with the AI. In the Gutenberg editor, drag in a Trilium AI Block, write your first prompt template, publish the page, and experience the power of AI workflow orchestration.
🔮 Looking Ahead: Limitless Possibilities in the Age of AI Agents
We’re at a thrilling inflection point. AI is evolving from a tool into a partner — shifting from passively answering your every question to proactively understanding your intent and taking action. This is the future that the AI agent wave represents, and Trilium AI stands at its forefront.
🌊 AI Agents: The Next Paradigm Shift
Over the past few years, we’ve witnessed the conversational AI revolution sparked by large language models and ChatGPT. But an industry consensus is forming: chat alone is just the tip of the AI iceberg.
The true next generation of AI applications will be agents that can plan autonomously, execute autonomously, and learn autonomously. They don’t just answer questions — they understand complex tasks, break them down into actionable steps, call on various tools to get the work done, and continuously accumulate experience along the way. From OpenAI to Google, from Anthropic to the open-source community, the entire AI industry is going all-in on agents.
Trilium AI’s Agent plugin was built for exactly this era. It lets AI agents operate on your knowledge base, execute multi-step workflows, and maintain memory across sessions — the defining traits of AI agents.
🐾 OpenClaw: A Rising Star in Agent Infrastructure
Within the AI agent ecosystem, OpenClaw represents a tremendously promising direction — local, decentralized, user-owned agent infrastructure.
Most AI agent solutions today are heavily dependent on cloud services. Your data, your conversation history, the AI’s “memory” of you — all stored on some company’s servers. OpenClaw offers a radically different path:
- 🏠 Your agents, running on your devices — No reliance on any centralized server
- 🧠 Your memory, stored on your hard drive — Full control over everything the AI knows about you
- 🔐 Your identity, protected by your keys — Device-level encrypted authentication, no third-party dependency
- 🔗 Open gateway architecture — Flexibly connect to various AI models under the hood, with no single-vendor lock-in
As the OpenClaw ecosystem matures, we can envision even more exciting possibilities: inter-agent collaboration, cross-device memory synchronization, community-shared agent templates, richer tool-calling capabilities… OpenClaw isn’t just building software — it’s building an AI agent platform that truly belongs to its users.
🗺️ Trilium AI’s Evolution Roadmap
As a deeply integrated member of the OpenClaw ecosystem, Trilium AI’s future is closely tied to the trajectory of AI agents. Here are the possibilities we’re exploring and envisioning:
🧠 Smarter knowledge understanding Future AI agents won’t just search your notes — they’ll understand the connections between them, automatically discovering conceptual links, identifying knowledge gaps, and suggesting content that needs to be added. Your Trilium knowledge base will evolve from a “note repository” into a “thinking knowledge graph.”
🤝 Inter-agent collaboration Multiple AI agents will no longer work in isolation. A writing agent finishes a draft and automatically hands it off to a proofreading agent; a research agent collects new information and proactively notifies the project management agent to update the timeline. Agent-to-agent collaboration will dramatically elevate the automation of knowledge work.
🧱 More powerful workflow orchestration Gutenberg Block workflow capabilities will continue to expand — supporting multi-step sequential execution, conditional branching, and data chaining between Blocks. One Block’s AI output can automatically become the next Block’s input, forming a true visual AI pipeline. Combined with WordPress’s page template system, you could build complete AI workstations tailored to different business scenarios.
📡 Cross-platform knowledge flow Trilium AI currently lives within the WordPress ecosystem, but knowledge shouldn’t be confined by platform boundaries. We plan to explore additional access points — browser extensions, desktop clients, mobile integrations — so your AI knowledge assistant is available wherever you are.
🎯 Proactive AI assistant Today’s AI agents still operate in “don’t ask, don’t tell” mode. The future direction is proactive AI — when it spots outdated information in your knowledge base, it alerts you to update it; when it notices that an article you’re writing is highly relevant to an old note, it proactively recommends it as a reference. Moving from “passive response” to “proactive assistance” is the next evolution of AI assistants.
🌐 A richer tool ecosystem OpenClaw’s open architecture provides limitless possibilities for tool extensions. Beyond Trilium Notes and FileBrowser, future integrations could include calendars, email, project management tools, code repositories, databases… The boundaries of what AI agents can do will keep expanding alongside the growing tool ecosystem.
💡 Why Trilium AI?
Among the sea of AI tools, what makes Trilium AI unique is that it’s not an isolated AI chat window, but a complete AI work system with your knowledge base at its core, agents as its engine, and WordPress as its orchestration platform:
🎯 Knowledge-driven, not conversation-driven — Most AI chat tools produce fleeting, isolated conversations. Trilium AI connects directly to your personal knowledge base, enabling AI to search, understand, and operate on notes and documents you’ve built up over years. Every conversation is grounded in your own knowledge.
🧱 WordPress-native workflow orchestration — Through the Gutenberg Block’s variable system and multi-instance capability, you can design AI workflows in the WordPress editor with zero code — from a simple translation assistant to a complex multi-step analysis pipeline, all via drag and drop.
🧠 Gets smarter the more you use it — Through OpenClaw’s persistent memory system, AI agents accumulate understanding of you across session boundaries. They don’t just remember what you said — over time, they increasingly grasp your work patterns, writing style, and thinking habits.
🏠 Your data, your rules — In an era of data anxiety, Trilium AI offers a genuinely local-first approach. The Ollama + OpenClaw + local FileBrowser stack lets you build a fully private AI system where not a single byte needs to leave your premises.
🔄 From conversation to automation — Trilium AI doesn’t stop at “chatting.” The workflow engine, multi-agent management, command system, and Gutenberg orchestration combine to automate your entire knowledge workflow — from information retrieval to content analysis.
🔮 Built for the agent era — Choosing Trilium AI means choosing not just a tool that works well today, but an ecosystem that evolves alongside the AI agent wave. As OpenClaw and AI agent technology advance, Trilium AI’s capabilities will keep growing with them.
🧑💻 About Us
Trilium AI is developed and maintained by SatoshiWP, released under the GPL v2+ open-source license. We believe the future of AI shouldn’t belong exclusively to tech giants — everyone deserves an AI assistant that is truly their own and truly understands them.
Trilium AI’s code is fully open, depends on no frontend frameworks, and stays lightweight while pursuing a best-in-class feature experience. We welcome community contributors to join us in building an AI knowledge engine that belongs to everyone.
💡 Trilium AI — Your notes deserve more than sitting as static text. They deserve an AI brain. Set your knowledge in motion, make AI an extension of your thinking, and let agents handle the tedious knowledge work for you. This isn’t a vision of the future — it’s a journey you can start right now.