An AI agent orchestration platform developed by KandaQuantum, a Japanese company. It controls multiple AI models in parallel and cooperation: describe your requirements in a format such as YAML, and the platform automatically generates deliverables including text, images, video, 3D models, and program code. Alongside fast code generation, it integrates a reverse-engineering feature that analyzes existing systems and visualizes their structure in 3D space, plus tools for managing the progress of multiple agents — aimed at developers and companies who want to automate the entire process from requirements definition through implementation.
Key Features
- Parallel, cooperative control of multiple AI models: Orchestrates multiple AI models such as the GPT series, Claude, and Gemini, having them generate deliverables in parallel against a single set of requirements. Handles large-scale generation tasks that are difficult with a single-model chat
- Automatic generation from written requirements: Describe requirements in structured text such as YAML, and the platform generates not only program code but also text, images, video, and 3D models. The workflow it aims for is “write the requirements, get the implementation”
- Fast code generation: Builds on the lineage of Zoltraak, KandaQuantum’s code-generation framework, enabling batch generation of many program files in a short time
- Reverse engineering in 3D space: Analyzes the source code of existing systems and visualizes their structure as a graph in 3D space. Zoom from a bird’s-eye view down to details to intuitively grasp the overall picture of complex systems
- Agent progress management: Built around the assumption that multiple AI agents run concurrently, it integrates tools for tracking each agent’s progress — supporting the operational side of agent-driven development
- Made in Japan: Developed by KandaQuantum in Japan, with documentation and communications primarily in Japanese, making it easy for domestic teams and companies to adopt
Pricing
A paid, subscription-based service with no free plan. According to third-party media, pricing is based on a monthly plan ($98/month) with discounts for 3-month and 6-month commitments, but the latest prices on the official site may have changed — be sure to check the official pricing page before subscribing.
| Plan | Price (reference) | Notes |
|---|---|---|
| Monthly | $98/month | Access to all features |
| 3-month plan | approx. $88.20/month | 10% discount |
| 6-month plan | approx. $78.40/month | 20% discount |
Prices are third-party reported figures as of August 2026. For the latest and accurate pricing, check the official pricing page.
Pros & Cons
✅ Pros
- Lets you experience an orchestration-style development workflow ahead of the curve: write requirements and multiple AIs generate deliverables in parallel
- Broad range of generation targets — not just code, but also images, video, and 3D models
- The 3D reverse-engineering visualization directly addresses a real-world problem: understanding the structure of legacy systems
- A Japan-made service with easy access to Japanese-language information and support
⚠️ Cons
- No free plan; at $98/month and up, pricing is on the high side for individuals
- Has its own conventions, such as writing requirements in YAML, so the learning curve is steeper than a general-purpose AI chat
- The field is evolving quickly; features and pricing can change fast, requiring you to track official information
- Compared with similar English-language services, there is less community knowledge and third-party documentation
Comparison with Similar Services
| Comparison | KAMUI | Devin | Claude Code | GitHub Copilot |
|---|---|---|---|---|
| Provider | KandaQuantum (Japan) | Cognition AI | Anthropic | GitHub / Microsoft |
| Form | Orchestration platform | Autonomous dev agent | Terminal-resident agent | In-editor assistant |
| Strengths | Parallel multi-AI control, 3D visualization, multi-format generation | Autonomous task execution | Codebase understanding, interactive development | Easy completion and chat |
| Reverse-engineering visualization | 3D spatial graph | None | None | None |
| Free plan | None | None | None (bundled with Claude Pro etc.) | Available |
Who Is It For
- Development teams that want to try a workflow where AI handles everything from requirements definition through implementation
- Companies that want to visualize and understand the structure of large, black-boxed legacy systems
- People who want to build generation pipelines combining multiple AI models and agents
- Organizations that prioritize a Japan-made service and Japanese-language information when choosing an AI development platform
Summary
KAMUI is an ambitious Japan-made AI agent platform that bundles parallel control of multiple AI models, automatic generation from requirements, and 3D reverse engineering into one product. Rather than a mere coding assistant, it aims at “automating the entire development process,” and features like legacy-system visualization address genuine practical needs. On the other hand, there is no free plan and a learning curve to climb — so check the latest official information and demos first, and make sure it fits your team’s workflow before adopting it.