Workshop Setup

Follow this checklist to get ready for the AI for TAMs workshop.

Goose Desktop Setup

Goose Desktop is a powerful AI agent that can analyze multiple files, reason across them, and generate complex outputs.

Installation

1
Install via OVH Self Service
macOS: Via OVH Self Service
Windows: Via OVH Catalogue
2
First launch
On first launch, the provider configuration wizard appears. Choose “Custom Provider” (last option).
3
Configure OVH Provider
Provider Type: OpenAI Compatible
API URL: https://oai.endpoints.preprod.ai.cloud.ovh.net/v1
API Key: Your OVH CORP email address
Models: Copy-paste the full list of OVH models (see below)

Available OVH Models

Copy this list into the “Available Models” field:

“id”: “DeepSeek-R1-Distill-Llama-70B”, “id”: “gpt-oss-120b”, “id”: “Qwen3Guard-Gen-0.6B”, “id”: “gpt-oss-20b”, “id”: “Llama-3.1-8B-Instruct”, “id”: “Qwen3Guard-Gen-8B”, “id”: “Qwen/Qwen3.5-397B-A17B”, “id”: “Mistral-Small-3.2-24B-Instruct-2506”, “id”: “ppl”, “id”: “Qwen3-32B”, “id”: “Qwen3-Coder-30B-A3B-Instruct”, “id”: “Mistral-7B-Instruct-v0.3”, “id”: “Mistral-Nemo-Instruct-2407”, “id”: “Qwen2.5-VL-72B-Instruct”, “id”: “Mixtral-8x7B-Instruct-v0.1”, “id”: “Meta-Llama-3_3-70B-Instruct”, “id”: “Qwen3-Embedding-8B”, “id”: “bge-multilingual-gemma2”, “id”: “bge-base-en-v1.5”, “id”: “bge-m3”, “id”: “whisper-large-v3-turbo”, “id”: “whisper-large-v3”, “id”: “stable-diffusion-xl-base-v10”, “id”: “stabilityai/stable-diffusion-xl-base-1.0”

Goose CLI Setup

Goose CLI is a command-line tool for automation tasks. Describe what you want in natural language, and Goose executes it.

Installation

Windows

1. Install WSL (Ubuntu or Debian)
2. Open your WSL terminal
3. Run the installation command below

macOS

1. Open Terminal (⌘+Space, type “Terminal”)
2. Run the installation command below

Installation command (WSL/macOS):

curl -LsSf https://github.com/block/goose/releases/latest/download/goose-x86_64-unknown-linux-gnu.tar.gz | tar xzf - -C ~/.cargo/bin

Environment Variables

Option 1: Quick (current session)

Run these commands directly in your terminal:

export GOOSE_TELEMETRY_ENABLED=false
export GOOSE_PROVIDER=custom_ai_endpoint
export GOOSE_MODEL=Qwen/Qwen3.5-397B-A17B
export GOOSE_MODE=approve
export CUSTOM_AI_ENDPOINT_API_KEY="your.email@corp.ovh.com"

⚠️ Variables will be lost when you close the terminal

Option 2: Permanent (recommended)

Add to your shell config file:

1. Open .bashrc or .zshrc
2. Paste the same export lines at the end
3. Run source .bashrc (or source .zshrc)

✅ Variables persist across terminal sessions

Test

goose version

🎁 Bonus: Editor Setup

Optional – enhance your workflow with these AI-powered editors.

VS Code Setup

Install VS Code from the OVH Catalog (Software Center / Self Service), then add these extensions:

AI Extensions

GitHub Copilot

Inline completions and chat panel. Install both:

• GitHub Copilot
• GitHub Copilot Chat

Kilo Code

Connect to OVHcloud AI Endpoint for internal/Orange Zone repos. Keeps all AI traffic on OVH infrastructure.

Claude Code

Full-agent coding with Claude. Requires Claude Team seat. Green Zone repos only (external infrastructure).

Recommended Extensions

GitLens

Enhanced Git history and blame.

Error Lens

Inline error/warning display.

Path Intellisense

File path autocompletion.

Rainbow CSV

Color-code CSV columns for easier analysis.

Cursor Setup

Cursor is an AI-first code editor with built-in chat, inline completions, and agent capabilities.

💡 Note: Cursor is a fork of VS Code—same interface, same extensions, same shortcuts. If you know VS Code, you already know Cursor.

1
Request via Arsenal
Go to the Arsenal form and submit your request.
2
Wait for approval
Once the task is resolved, your access is granted.
3
Log in
Open Cursor and log in with firstname.lastname@corp.ovh.com.

SHAI Setup

SHAI is a CLI-based coding agent that integrates with OVHcloud AI Endpoints.

Installation

Follow the official SHAI README to install the CLI.

Configure Provider

Run shai auth and configure:

• Provider: ovhcloud
• OVH_BASE_URL: https://oai.endpoints.preprod.ai.cloud.ovh.net/v1
• OVH_API_KEY: your professional email (firstname.lastname@corp.ovh.com)
• Model: choose your preferred model (e.g., Qwen/Qwen3.5-397B-A17B)

Important

The internal AI Endpoints configuration only works when connected to the OVHcloud internal network (offices/datacenters) or via VPN.

Usage

Type shai to launch the agent and start coding with AI Endpoints! 🚀

Questions?

If you encounter any issues during setup, reach out to Yann Dabout directly.