32dots HEIDELBERG AI
Curriculum › Local & Private Models › LM Studio course
EXTENDED COURSE

LM Studio

Five hands-on lessons take you from installing LM Studio and picking your first model through running a local API server and wiring it to Hermes — so your personal AI assistant runs fully offline on your own hardware. This course doubles as the local-model setup for the Hermes course.

1Lessons5step-by-step, ~90 min each

Setup

  1. 00 Install LM Studio & download your first model

Basics

  1. 01 Chat locally & compare models side by side
  2. 02 Turn it into a local API server

Advanced

  1. 03 Power Hermes with your local model
  2. 04 Run it headless / always-on
2Cheat sheetcopy-ready expressions
3Examples4what people built

Internal tools & ops1

LM Studio AI-native SME Small biz
In the gallery

Contract clause review without a legal API

A business owner pastes supplier or client contract text into LM Studio's chat and asks the model to flag unusual indemnity clauses, payment terms, and jurisdiction choices — all processed locally before involving a lawyer for final sign-off.

A first-pass legal screen is done in minutes on sensitive contract text, without uploading confidential terms to a cloud AI service.

Try it yourself

In LM Studio, use a large instruction model. System prompt: "You are a contract analyst. Flag clauses that are unusually risky for the buyer. Be specific and cite the clause text." Paste the contract.

Content & marketing1

LM Studio Scientist
Course starter

Draft grant application sections offline

A researcher uses LM Studio's chat interface on a field laptop without internet to draft specific grant sections — significance, innovation, approach — by feeding the model their notes and asking for structured prose that matches funder guidelines.

Grant drafts progress during travel or remote fieldwork with no connectivity dependency and no risk of unpublished ideas reaching a cloud provider.

Try it yourself

In LM Studio, load a strong instruction model (e.g. Qwen 2.5 14B), paste your bullet-point notes, and prompt: "Write the Significance section of an NIH R01 application based on these notes. Use formal scientific register."

Research & data tools1

LM Studio Scientist
Course starter

Q&A over unpublished patient notes (no cloud)

A researcher loads a folder of de-identified clinical notes into LM Studio's local OpenAI-compatible server, then queries it in a simple chat UI to extract specific findings — adverse events, dosing records, outcome dates — without any data touching a cloud endpoint.

Ethics-compliant analysis of identifiable data is possible on a standard laptop, even before IRB-approved anonymisation is complete.

Try it yourself

In LM Studio, download a medical-tuned model (e.g. Meditron or a Mistral variant), start the local server, then chat: "From the notes below, extract all adverse events and their dates." Paste notes into the message.

Knowledge & docs1

LM Studio AI-native SME
Course starter

Private HR policy chatbot for an SME

HR uploads the staff handbook and policy PDFs to a local folder; LM Studio's chat UI is used with a system prompt grounding answers to that folder's content — employees ask questions about leave, expenses, and conduct in a local web interface with no data leaving the office network.

Routine HR queries are answered instantly without involving a SaaS chatbot that would hold proprietary policy text on external servers.

Try it yourself

In LM Studio, load a mid-size instruction model, set system prompt: "Answer only from the provided policy documents. If unsure, say so." Paste the relevant policy text into the system context or use LM Studio's document-grounding feature.