Add local AI coding to your professional toolkit.
Generate PowerShell scripts, troubleshoot systems, document infrastructure—all without sending data to the cloud.
You're an IT professional—support, network, infrastructure, maybe development. You have deep expertise in your domain. This guide helps you add a new capability: running powerful AI models locally for daily work. Generate automation scripts, troubleshoot error messages, write documentation. Practical tasks that justify the setup time.
What makes this different: No cloud dependencies, no data leaving your organization, no monthly subscriptions. Your hardware, your models, your data.
LM Studio handles model discovery, download, and serving. Intuitive CLI, robust functionality, chat GUI with tool calling.
opencode provides the coding workflow. Connect it to LM Studio's served models and you have a complete local AI coding environment.
Target configuration: 32GB RAM systems (Ryzen AI, Intel Lunar Lake) running Qwen 3.6 35B or similar models via Vulkan.
16GB systems: See sidebars throughout the documentation for 9B model alternatives that work within tighter memory constraints.
graph LR
A[You] -->|Commands| B[opencode]
B -->|API Calls| C[LM Studio Server]
C -->|Loads & Runs| D[AI Model]
D -->|Responses| B
style B fill:#7C3AED,color:#fff
style C fill:#0078D4,color:#fff
style D fill:#00ADD8,color:#fff
Key constraints for 32GB:
- Context: 32,768 tokens
- Concurrency: 1 request at a time
- Single model: Unload before loading another (RAM physics, not ceremony)
These are tested defaults that keep everything stable when running other software alongside your AI tools.
| Document | When to Read |
|---|---|
| QUICKSTART | Start here. Zero to running model in minimal steps. |
| SETUP | After QUICKSTART works and you want to understand what you built. |
| CONFIG | When customizing model settings or switching between models. |
| CAVEATS | Reality check: honest assessment of tradeoffs, costs, and limitations. |
| NOTES | For design rationale, hardware recommendations, troubleshooting. |
| Use Cases | See it in action: PowerShell generation and error troubleshooting. |
Are you familiar with local AI setup already? Then you can use these three commands to get running:
# Install LM Studio
winget install ElementLabs.LMStudio --accept-package-agreements --accept-source-agreements
# Download and start
lms get <model-name>
lms load <model-name> --context-length 32768 --parallel-requests 1DO you need step-by-step guidance? Then follow QUICKSTART.md for the complete setup sequence.
Primary: Field IT Pros learning to add AI coding to their toolkit. You might not have touched git or developer workflows before—that's fine. We explain what matters without assuming developer background.
Also useful for: Developers (who already know opencode), Infrastructure Architects, Software Architects, Specialists/SMEs, IT Team Managers.
If you're a professional looking to expand your capabilities with practical AI tools, this documentation is for you.
