AI-Powered IDE Multi-Provider

Your AI Assistant
for Every Task

Tao AI Studio is a desktop IDE that orchestrates multiple AI models simultaneously — with persistent memory, parallel agents, document editing, and tool calling built in from the ground up. It ships as a single EXE with no complex installation — just .NET 9 is required, which comes standard with Windows 11.

Windows 10/11 · .NET 9 Runtime included · Free

5+
AI Providers
14
Tab Types
100
Root Agent Turns
6-Step
Memory Retrieval
Tao AI Studio — main-window.xaml
hero-screenshot.png
Ollama — Local LLMs
OpenRouter — Cloud Models
vLLM — High-Throughput Server
LocalAI — Self-Hosted
Hugging Face — Inference API
Llama.cpp — Direct Server
Ollama — Local LLMs
OpenRouter — Cloud Models
vLLM — High-Throughput Server
LocalAI — Self-Hosted
Hugging Face — Inference API
Llama.cpp — Direct Server

Everything you need in one IDE

From code editing and AI chat to office document generation and autonomous agents — all in a single desktop application.

🧠

Memory Palace

Persistent episodic memory with vector search. Memories are organised into Rooms, Halls, Chains, and Tunnels — surfacing the right context every time.

Parallel Agents

Spawn independent background worker agents that run up to 40-turn agentic loops concurrently. Monitor live progress without blocking the main chat.

📊

AI-Powered Spreadsheet

Edit spreadsheets with natural language. Type "Create a budget tracker" and watch the AI generate formulas, data, and formatting automatically.

🎯

Goal Check

Automatic self-review loop that keeps agents accountable. The agent evaluates its own work against project tracker goals and continues until satisfied.

🔧

Smart Tool System

200+ available tools organised into categories that auto-expire after 10 seconds of inactivity, keeping every LLM call lean and focused.

🌐

Multi-Provider Routing

Switch between Ollama, OpenRouter, vLLM, LocalAI, HuggingFace, and Llama.cpp with a model prefix. Unified message format across all providers.

📝

Code Editor

Full-featured AvalonEdit integration with syntax highlighting, multi-tab support, and automatic file type detection across 15+ languages.

🗺️

Mermaid Diagrams

Generate class, sequence, component, and namespace summary diagrams. Pre-generated diagrams can be injected into the system prompt for architectural awareness.

👁️

Vision & Image Gen

Vision model support for image analysis plus text-to-image generation. Attach images, videos, audio, and documents as conversation context.

🧊

2D → 3D Conversion

Transform any image into a 3D model using HuggingFace Trellis or Stability AI's Stable Fast 3D. Output GLB files and view them in the built-in Three.js 3D viewer tab.

🔌

MCP Support

Model Context Protocol integration — connect any MCP server and its tools are automatically discovered and merged with the built-in tool library.

💡

Extended Thinking

Enable reasoning mode on supported models with a configurable 1K–32K token budget. Works with OpenRouter thinking models, vLLM, and llama.cpp /think commands.

🗂️

Plan Mode

A structured planning interface lets the agent outline its approach before executing. Working memory gives agents a persistent scratch pad across turns.

🌐

Brave Web Search

Integrated Brave Search API gives agents real-time internet access with 2,000 free calls/month. Paired with a web-fetch tool for full page content extraction.

📧

Outlook Integration

9 email and calendar tools — read unread mail, send emails with attachments, create appointments, search inbox, and manage contacts via Microsoft Outlook.

One workspace, nine tab types

Open any file and the right editor loads automatically. Every document type gets its own AI-integrated experience.

📄

File Editor

AvalonEdit code editor with syntax highlighting for 15+ languages

💬

AI Chat

WebView2 chat with markdown rendering and streaming responses

🤖

Agent Chat

100-turn autonomous agent with full tool access and Goal Check

🖼️

Image Viewer

Display generated and processed images; attach to chat for vision analysis

🎬

Video Player

Playback MP4, AVI, MOV, MKV, WEBM directly in a tab

🎵

Audio Player

WAV, MP3, M4A, OGG, FLAC playback plus AI speech synthesis output

📊

Spreadsheet + AI

DataGrid spreadsheet with inline AI chat for natural language editing

📝

Document + AI

Document tab with integrated AI chat for content generation

📑

Presentation + AI

Slides creation and editing with AI chat assistance

📰

PDF Viewer

View and interact with PDF documents with integrated AI chat

🗺️

Mermaid Viewer

Rendered Mermaid diagrams — class, sequence, component, namespace

🧊

3D Viewer

Three.js GLB viewer with pan, zoom, and rotate — outputs from Trellis/Stable Fast 3D

🌐

Web Page

Embedded browser tab with back/forward/refresh navigation controls

🎨

SVG Viewer

Render and inspect SVG drawings and vector graphics inline

tab-system-overview.png

Open any file, get the right experience

Tao AI Studio automatically detects file type on open and routes to the correct tab — code files open in AvalonEdit, spreadsheets launch the Spreadsheet+AI tab, images appear in the viewer, and media files play immediately.

  • Syntax highlighting for all major languages via AvalonEdit
  • Automatic file type detection from extension and content
  • AI integrated into every Office document tab
  • Speech synthesis for AI responses in any tab

Edit as text. Preview with one click.

Open any .html or .md file as raw text in the code editor. Click the filename or the eye icon in the tab bar and the file instantly opens in its dedicated preview tab — no manual conversion needed.

HTML → Web Page Tab Web Page
preview-html-to-webpage.png

Edit .html source in the code editor. Click the filename or eye icon — a Web Page tab opens with the rendered site, full navigation controls included.

Markdown → Preview Tab Markdown Preview
preview-md-to-markdown.png

Edit .md source in the code editor. Click the filename or eye icon — a Markdown Preview tab opens with fully rendered headings, code blocks, and tables.

Run any model,
anywhere

Switch between local servers and cloud APIs with a simple model prefix. A single unified message format keeps your code provider-agnostic.

🦙

Ollama

Local LLM server with native concurrent client support. The default provider — zero config needed for most setups.

No prefix / OL --
🌐

OpenRouter

Cloud gateway to hundreds of frontier models. Real-time cost and token tracking per message. Extended thinking supported with 1K–32K token budgets.

OR --

vLLM

High-throughput inference server. Configurable endpoint in settings — ideal for quantised open-weight models at scale.

vLLM --
🏠

LocalAI

Self-hosted OpenAI-compatible server. Single-process — the agent semaphore serialises worker calls automatically.

LocalAI --
🤗

Hugging Face

Inference API integration supporting thousands of models on HF Hub. Requires an API key configured in Settings.

HF --
🦙

Llama.cpp

Direct llama.cpp server integration for maximum control over quantisation and GPU offloading on consumer hardware.

LL --
// Provider routing — zero boilerplate in your prompts
var model = "OR -- anthropic/claude-3-5-sonnet"; // → OpenRouter
var model = "HF -- meta-llama/Llama-3-8B-Instruct"; // → HuggingFace
var model = "vLLM -- mistral-7b-instruct"; // → vLLM server
var model = "qwen2.5-coder:7b"; // → Ollama (default)

AI memory that thinks like you do

Inspired by the Method of Loci, the Memory Palace organises knowledge spatially. Every conversation is encoded, clustered, linked, and retrievable in milliseconds — with zero extra VRAM.

  • Rooms — topic clusters formed by cosine centroid (threshold 0.58)
  • Halls — knowledge types: fact, procedure, event, decision
  • Chains — sequential episode links for conversation context
  • Tunnels — cross-topic bridges (threshold 0.65)
  • RAM-first — ConcurrentDictionary for instant recall, LiteDB for persistence
  • Three embedding modes — CurrentModel, DedicatedModel, or Keyword TF-IDF
Cosine Similarity Salience Boost Recency Boost LLM Re-ranking
1
Room
Detection
2
Scoped
Search
3
Filter by
Hall / Date
4
Chain &
Tunnel Expand
5
Token
Budget Trim
6
LLM
Re-rank
🏠 Room

Auto-clustered topic groups. Each room has a generated label from anchor keywords.

🏛️ Hall

Knowledge classification: fact · procedure · event · decision

⛓️ Chain

Episode links via PreviousMemoryId — walk the chain for full conversation replay.

🚇 Tunnel

Cross-topic bridges at 0.65 similarity — surface related ideas from other projects.

memory-palace-explorer.png

Zero extra VRAM required

The default CurrentModel strategy reuses the active chat model for embeddings — no second model to load, no extra GPU memory consumed. Switch to a dedicated encoder for higher precision, or to keyword TF-IDF for fully offline operation.

  • CurrentModel — zero extra VRAM, uses chat model's /v1/embeddings endpoint
  • DedicatedModel — separate encoder, e.g. nomic-embed-text
  • Keyword — pure TF-IDF hash, no API calls at all

Spawn workers. Stay in flow.

The root agent runs up to 100 turns autonomously and can spawn background workers — each running their own independent 40-turn agentic loop. The root never blocks while workers run in parallel.

Root Agent 100 turns

  • 100-turn autonomous agentic loop
  • Spawns workers via launch_parallel_agent
  • Polls worker status any time
  • Retrieves completed results
  • Can kill workers if needed
Task.Run Poll
Workers — 40 turns each
worker-research Turn 12/40 · Brave Search
worker-codegen Turn 27/40 · Writing tests
worker-analysis Completed · Results ready
parallel-agent-status-dialog.png

Live progress without polling overhead

Each worker writes its state to .tao/agents/{name}/. The AgentStatusDialog uses a FileSystemWatcher for instant OS-level notifications — no polling, no timers.

  • config.json — worker configuration and model choice
  • status.json — live progress, current tool, turn count
  • progress.log — full tool call history
  • results.json — final output when complete

Concurrent without race conditions

Each provider gets a dedicated service instance per worker — no shared event handlers, no LLM response races. A single semaphore serialises calls to single-process servers like vLLM and LocalAI.

  • Dedicated instances per worker per provider
  • Semaphore queue for single-process servers
  • Ollama handles concurrency natively
  • OpenRouter is cloud — no VRAM constraint

Agents that verify themselves

Enable Goal Check and the agent is held accountable to your active Project Tracker goal. When it stops, it must evaluate its own work — and keeps going until it's genuinely satisfied.

1

Agent stops naturally

No tool calls detected — the agent appears to be done with its current response.

2

Self-review injected automatically

A structured review message referencing the active goal is injected as a user message — covering quality, testing, output verification, and blockers.

3

Agent calls completed()

satisfied: true — stop. satisfied: false — continue autonomously until the next review.

4

Loop continues until done

The cycle repeats each time the agent stops — guaranteeing it doesn't declare victory prematurely.

goal-check-review-flow.png
Review Criteria
Goal completion verified
All sub-tasks broken down & done
Output quality reviewed
Tests written & run
Screenshots taken for visual output
Project tracker fully updated
Hard blockers reported to user

Documents powered by AI agents

The agent can autonomously create, populate, verify, and screenshot Office documents — all via tool calls, with tab locking to prevent interference.

Spreadsheet AI Workflow
💬
1

Natural Language

"Create a monthly budget tracker with formulas"

⚙️
2

LLM Processing

Current spreadsheet state + message → JSON operations from LLM

📊
3

ClosedXML Apply

State manager applies cell edits, formulas, and formatting

🔄
4

UI Refresh

DataGrid updates instantly — iterate with follow-up messages

excel-ai-chat.png

Agents build documents autonomously

The OfficeTabs tool category lets agents open, command, and verify Office documents. Tab locking prevents the user from interfering while the agent works.

open_office_tab("Spreadsheet") → tab_id
office_command(id, command) → state snapshot
screenshot_tab(id, path) → PNG for verification
focus_tab(id) → foreground
  • Tab locking blocks user navigation while agent works
  • Programmatic screenshots — no focus required
  • Status bar indicator shows amber lock message
  • Works with spreadsheets, documents, and presentations
word-powerpoint-ai.png

Every Office format, AI-assisted

Agents can create polished documents and presentations from scratch. Each tab includes an embedded AI chat — ask follow-up questions, request edits, or have the agent iterate on content.

  • Integrated AI chat within every document tab
  • Document-specific system prompts via .tao folder
  • JSON command protocol for precise edits
  • Screenshot verification before task completion

Image in. 3D model out.

Drop any 2D image and Tao AI Studio converts it to a production-ready 3D GLB model via HuggingFace Gradio Spaces — no local GPU required. The result opens instantly in the built-in Three.js viewer.

🔮
HF Trellis

JeffreyXiang/TRELLIS — highest quality multi-view 3D reconstruction. Best for objects and products.

Stable Fast 3D

stabilityai/stable-fast-3d — ~30 second conversion via Stability AI's HuggingFace Space. Fast previews.

  • Full Gradio REST API — upload → submit → SSE poll → download
  • GLB output — standard binary format compatible with all 3D tools
  • Built-in 3D viewer — pan, zoom, rotate with Three.js, no extra software
  • Cloud-based — uses HuggingFace Spaces, no local GPU needed
threejs-3d-viewer.png

120+ tools, zero bloat

Tool categories auto-disable after 10 seconds of inactivity. Every LLM turn gets only the tools it needs — keeping context windows lean and costs down. Connect MCP servers to add more.

⚙️

General

Core tools always available: completed(), enable_tool_category, working memory, plan mode, and agent management.

📁

File System

Read, write, glob, zip, hash. Directory ops, file moves, and properties. Auto-expires 10s after last use.

🌐

Web

Brave Search API, web fetch with content extraction, URL accessibility checks, and browser bookmarks.

💻

PC Control

Launch apps, screen capture, keystrokes, mouse control, clipboard, registry, network, system monitoring.

📊

OfficeTabs

Open and control spreadsheet, document, and presentation tabs. Screenshot verification included.

🗂️

Project Tracker

Goals, tasks, subtasks — plus pt_git_* tools for commit, push, diff, log, and branch from within the tracker.

🧠

Memory

Search by topic, date, or recency. Retrieve statistics and store new episodic memories mid-conversation.

🤖

Agents

Launch, poll, retrieve output from, and kill parallel worker agents. Root-agent only.

📧

Outlook

Read unread mail, send with attachments, create calendar appointments, search inbox, manage contacts.

🔌

MCP (Dynamic)

Tools from any connected Model Context Protocol server are auto-discovered and added to the active set.

🧊

3D Generation

Convert 2D images to GLB models via Trellis or Stable Fast 3D. Open output in the built-in Three.js viewer.

👁️

Vision

Inject images into the next turn with see_image, capture screen regions, and screenshot any open tab.

// Enable a category and it stays active for 10 seconds
enable_tool_category("OfficeTabs", true);
// … do office work …
// Category auto-expires — next turn is lean again

Production-grade open-source stack

Built on .NET 9 WPF with battle-tested libraries for every capability.

✏️

AvalonEdit 6.3

Full-featured code editor component with syntax highlighting, code folding, and line numbers.

🌐

WebView2

Microsoft's Chromium-based WebView for rendering AI chat with full markdown and HTML support.

📊

ClosedXML 0.105

Pure .NET spreadsheet library for creating and modifying spreadsheets — no Office installation required.

🗃️

LiteDB 5

Embedded document database for persistent memory storage — zero config, serverless.

🦙

OllamaSharp 5.4

Official .NET client for Ollama with streaming, tool calling, and model management.

🗣️

System.Speech

.NET speech synthesis for AI response audio playback — text-to-speech built into .NET 9.

.NET 9 · WPF · Windows

Build in Debug for development or Release for distribution. Open in Visual Studio for the full WPF design experience.

# Build and run
dotnet build TaoAIStudio.sln
dotnet run --project TaoAIStudio.csproj
🚀
Tao AI Studio — Coming Soon! We'll announce the release here when it's ready.