What We Do
We build local Windows tools that help you work with AI productively - without losing control of your data, your projects, or your budget.
The Problem
Most people use AI like a very long chat. You ask a question, get an answer, ask another, and another. Over time the conversation becomes huge. The AI starts forgetting details, mixing up old ideas with new ones, and drifting away from what you actually wanted. On real projects - writing software, planning architecture, managing tasks - this becomes expensive and frustrating.
You also have to trust cloud services with your ideas, your files, and sometimes your company data.
Our Solution
We split AI work into clear stages and keep it on your own machine. You get the power of modern AI models without the chaos of an endless chat; with local models, your data does not have to go to the cloud for that.
SindByte MCP Server
The local engine. It connects your AI host (like LM Studio or VS Code) to real Windows actions: file operations, desktop automation, image editing, and more. Runs on 127.0.0.1:5555 - on your machine.
WorkJinn
The project workspace. Instead of one giant chat, WorkJinn breaks projects into steps: set a goal, build an expert board, run action cycles, validate results, and hand off cleanly. Keeps projects on track even when they get large.
Download WorkJinnPlanJinn
The planning workspace. You describe a program, and PlanJinn turns it into a verified plan, lets the AI write the code piece by piece, and builds it with the built-in PBXB64 compiler. Helps you think before you build. Available now.
Download PlanJinnCompiler Workshop
Our own compiler line for Windows. The PBX Compiler V10 package contains PBXB64 (signed main compiler with native x64 output), ScharffBasic V1 Beta, PBXA64, PBXA32 and the PBXC64 preview. For developers who want low-level control and native performance.
Download PBX Compiler V10 (355 MB)Why Local?
Privacy
Your prompts, your files, your projects: with local models they stay on your computer. You only use cloud services and cloud AI hosts if you choose them yourself.
License and Costs
Free for private, non-commercial use. Business, team, and commercial use is licensed through Smart Package Robot (License explained). Local models have no API fees.
Speed
Local models, for example in LM Studio, run without a network trip to a provider and without its rate limits. How fast they are depends on your hardware.
Control
You decide which tools are published, which categories are visible, and which credentials are required. Short and Full registration modes are included.
How It Works
Download and start
Download SindByte.zip, extract it, and run SindByte 01.exe. The MCP endpoint is then reachable on port 5555.
Connect your AI
In LM Studio, Roo Code, Codex, or VS Code, enter the endpoint http://127.0.0.1:5555/mcp. Which tools appear depends on your filters and credentials.
Work productively
Use WorkJinn for structured projects, PlanJinn for planning, or call tools directly from your AI host. With a local model, all of this runs on your machine.
Ready to start?
Check the README and licence supplied with each download. Download SindByte and explore the AI Workshop. Business, team, and commercial use goes through Smart Package Robot or by email to info@it-berater.org (License explained).
See downloads and licence details AI Workshop Free intro call