Self-hosted AI coding assistants in AI Developer Productivity Tools | Parse

To run a fully local, private AI coding assistant on your laptop with zero telemetry or code snippets leaving your machine, the gold-standard open-source stack pairs Ollama (as the local inference engine) with Continue (as the IDE extension).

This setup provides autocomplete, side-panel chat, and codebase indexing entirely offline.

Ollama (The Local Engine)
- What it does: Runs large language models directly on your CPU or GPU locally.
- Why it fits: Operates completely offline, requiring no account creation or cloud connection.
- Get it at:
Continue (The IDE Extension)
- What it does: An open-source autopilot/chat extension for Visual Studio Code and JetBrains that connects directly to local backends like Ollama.
- Why it fits: Replaces cloud extensions (like GitHub Copilot) with a local UI that handles sidebar chat, inline code generation, and highlighting.
- Get it at: Continue.dev [](https://www.sitepoint.com/local-ai-coding-assistant-vscode-ollama-continue/)[[2](https://www.youtube.com/watch?v=Z1Yk83j6cyg)](https://www.youtube.com/watch?v=Z1Yk83j6cyg)

Step 1: Install and Run Ollama

Download and install Ollama for your operating system.
Open your terminal/command prompt and pull a strong, lightweight coding model. For a standard laptop, models optimized for code work exceptionally well:
- DeepSeek-Coder-V2 or Qwen2.5-Coder (7B or 1.5B parameters depending on your RAM/VRAM).
- Run this command in your terminal to download and run a coder model:bash``` ollama run qwen2.5-coder:7b

Use code with caution.
3. Once downloaded, Ollama runs a local server automatically in the background on `http://localhost:11434`.

#### Step 2: Install the Continue Extension

1. Open **VS Code** (or your JetBrains IDE).
2. Go to the Extensions marketplace, search for **Continue** , and install it.
3. Click the new **Continue icon** in your sidebar.[[1](https://www.sitepoint.com/local-ai-coding-assistant-vscode-ollama-continue)](https://www.sitepoint.com/local-ai-coding-assistant-vscode-ollama-continue/)[[2](https://www.youtube.com/watch?v=Z1Yk83j6cyg)](https://www.youtube.com/watch?v=Z1Yk83j6cyg)

#### Step 3: Configure Continue to Point to Ollama

1. Click the gear/settings icon at the bottom right of the Continue sidebar, which opens `config.json`.
2. Configure your models to use your local Ollama instance. A basic configuration looks like this:json```
{
  "models": [
    {
      "title": "Qwen2.5 Coder 7B",
      "provider": "ollama",
      "model": "qwen2.5-coder:7b"
    }
  ],
  "tabAutocompleteModel": {
    "title": "Qwen2.5 Coder 1.5B",
    "provider": "ollama",
    "model": "qwen2.5-coder:1.5b"
  }
}

Use code with caution.(Using a smaller 1.5B model for tab-autocomplete keeps inline suggestions snappy, while the 7B model handles heavy lifting in the chat panel). [](https://www.youtube.com/watch?v=Z1Yk83j6cyg)[[2](https://www.youtube.com/watch?v=Z1Yk83j6cyg)](https://www.sitepoint.com/local-ai-coding-assistant-vscode-ollama-continue/)

Would you like help choosing the ideal model size for your specific laptop's hardware specifications (RAM/GPU), or do you need help setting up tab-autocomplete specifically?

Sources cited in this answer2

Yes. If self-hosted + no telemetry is the priority, I’d look at these:

continue.dev — probably the easiest if you want a VS Code/JetBrains assistant. It supports local models, and its documentation explicitly describes an offline/air-gapped setup where you disable anonymous telemetry and use a local model.
tabbyml.com — a dedicated self-hosted coding-assistant server. It's open source, can run on your own hardware, and supports IDE integrations and code completion/chat.
aider.chat — excellent if you're comfortable with a terminal-based workflow. It can use entirely local LLMs, maps your codebase, and integrates with Git.
AIDE — a newer locally executed developer tool. Its current privacy policy states that it has no backend, accounts, analytics, telemetry, usage reporting, or crash reporting, with state kept on your filesystem.

If you mean literally zero network telemetry

I'd use Continue + a local model and configure it for offline operation. Continue's documentation specifically says that disabling anonymous telemetry prevents requests to PostHog, and with a local model your code/context can remain on the machine.

A typical private stack would be:

VS Code / JetBrains
        │
     Continue
        │
   local model
 (Ollama/llama.cpp/etc.)
        │
      GPU/CPU

For a stricter environment, you can also block the assistant's outbound network access entirely with a firewall/container network policy. That gives you a stronger guarantee than relying solely on an application's telemetry setting.

If you tell me your OS + GPU/RAM (e.g. Windows + RTX 4070 12 GB, MacBook M-series, Linux + 64 GB RAM), I can recommend a specific setup and model that will run locally.

Sources cited in this answer10

Tool Stack	Primary Use Case	Telemetry Status	Best Suited For
Ollama + Continue	Individual Developer / Homelab	Opt-out (Simple config toggle)	Quick 5-minute setup in VS Code.
Tabby	Team / Centralized Server	Opt-out / Supports fully air-gapped	Multi-user setups with repository context.
VSCodium + Roo Code	Strict Privacy Purists	Zero Telemetry by default in IDE	Autonomous agent workflows without cloud leaks.

Self-hosted AI coding assistants

Which brands AI mentions

Recommendations by AI engine

Questions and answers

My top recommendations

1. Continue + Ollama

See where your brand stands in this ranking

2. Aider + Ollama

3. Tabby

How to make sure your code stays private

Let me narrow it down to the best setup for your laptop