Sébastien Burel
Sébastien Burel is an individual software engineer, entrepreneur and solo creator rather than a standalone company. Through Haruni, he designs digital products and client applications focused on generative AI, cloud architecture, mobile and web development; his current product work is TyKaoz, a private, local-first AI wiki for Mac whose open-source foundation is KaozKit.
At a Glance
- Individuals and teams needing private local document search and AI answers
- Mac developers building native Swift applications with embedded agents
- Startups and SMBs needing AI architecture or CTO guidance
- Organizations deploying generative-AI, RAG or multi-model systems
- +1 more
AI Tools by Sébastien Burel
(1)
KaozKit
JS LLM Agents in Swift Apps
Discussions
No discussions yet
Be the first to start a discussion about Sébastien Burel
Latest News
KaozKit 0.2.0 released
Burel published 'A JavaScript engine built for microcontrollers turns out to be a great runtime for AI agents,' explaining KaozKit and its role as the foundation of TyKaoz.
KaozKit 0.1.1 released as an early public release for macOS 26+ and Apple Silicon.
Haruni published the KaozKit/XS/Swift article describing stateful AI agents, heap snapshots and TyKaoz's local-first vision.
Products & Services
A private, local-first smart wiki for macOS that answers questions from the user's own documents with sources. It is designed so documents, queries and conversations remain on the Mac; it supports cloud models such as Claude, GPT and Mistral as well as local models.
An MIT-licensed Swift package for autonomous LLM agents written in JavaScript and embedded in Swift apps using Moddable's XS engine. It includes the agent runtime, providers, tools, memory, channels and persona support.
The reusable JavaScript-to-Swift engine layer within KaozKit: an XS engine bridge with asynchronous calls, streaming and heap snapshot/restore support.
An optional KaozKit product layer providing on-device MLX inference providers.
Market Position
Burel positions TyKaoz against cloud-first AI assistants by keeping documents, indexes, queries and conversations on the Mac. KaozKit differentiates from heavier server runtimes such as Node or Python by embedding a small, confined XS JavaScript engine in Swift, with restartable heap snapshots, native async bridging and on-device providers; the trade-off is its narrow platform target (macOS 26+ and Apple Silicon) and pre-1.0 maturity.
Founding Story
Burel says he has been building digital products for more than 20 years. He launched Haruni around 2005 as a legal/business home for his work, after earlier technical and entrepreneurial experience; later he co-created Frigo Magic in 2015 to turn ingredients people already had at home into recipe suggestions. His current TyKaoz effort began as a response to the need for a private AI wiki in which documents and conversations remain on the user's Mac, leading him to build KaozKit as its agent-runtime foundation.
Business Model
Revenue Model
Burel's Haruni practice generates revenue through remote consulting and software-development engagements covering generative AI, AWS cloud, mobile/web development and architecture/CTO-as-a-service. TyKaoz is planned as a paid macOS license product, with payments handled by Paddle; it offers a free 14-day trial and a 14-day refund window.
Pricing Tiers
No account, card or sign-up required, according to the official refund policy.
Target Markets
- Individuals and teams needing private local document search and AI answers
- Mac developers building native Swift applications with embedded agents
- Startups and SMBs needing AI architecture or CTO guidance
- Organizations deploying generative-AI, RAG or multi-model systems
- Clients needing AWS cloud, iOS, Android or web product development
- Private personal or small-team knowledge wikis over local documents
- On-device AI assistants where sensitive files and conversations should stay on a Mac
- Stateful, restartable autonomous agents in native Swift applications
- RAG, autonomous agents, fine-tuning and multi-model orchestration
- AWS cloud architecture and deployment for AI applications
- Mobile and web product development from MVP through production