AI Model Service Account — Setup Guide

Overview Use this guide to create and configure an AI Model service account in MinuteView Console. This is the single account type that provides Large Language Model (LLM) access for all of MinuteView's AI features:

• AI-powered search in Mesh, including the search and chat agents
• PowerShell Code Assistant autocomplete and inline help
• Automation workflows with AI/natural-language nodes
• Document analysis, summarisation, and chat assistants

Each AI Model account connects to one provider. You choose the provider, enter that provider's API key, and specify a model ID. Create multiple accounts if you want to mix providers or models (e.g. a fast/cheap account and a high-reasoning account).

Replaces the old "OpenAI" account

Earlier versions had a dedicated OpenAI service account. That has been replaced by this provider-agnostic AI Model account, which supports OpenAI, Anthropic (Claude), and Google (Gemini). If you previously used an OpenAI account, create an AI Model account with Provider = OpenAI and the same key.

---

Network Requirements

Although this account authenticates with an API key (not an OAuth login), MinuteView calls the AI provider server-side — the key is sent with every request from the MinuteView server and, for AI workflow nodes, from the automation engine. The server must therefore have outbound internet access to the host for the provider you select.

Required Firewall / Proxy Whitelist

Whitelist the host for your chosen Provider, accessible via HTTPS (port 443) from the server where MinuteView is installed:

Provider	Domain
OpenAI	api.openai.com
Anthropic (Claude)	api.anthropic.com
Google (Gemini)	generativelanguage.googleapis.com

Custom or gateway endpoints

These are the default hosts. If your account is pointed at a custom Base URL (an Azure OpenAI resource, a proxy, or an internal AI gateway), whitelist that host instead. If the server cannot reach the configured host, AI features will fail even though the key is valid.

---

1) Get an API Key for Your Provider

Pick the provider you want this account to use, then obtain a key:

OpenAI

1. Sign in at https://platform.openai.com/
2. Go to API Keys → Create new secret key
3. Copy the key (sk-proj-... or sk-...) — it is only shown once

Anthropic (Claude)

1. Sign in at https://console.anthropic.com/
2. Go to API Keys → Create Key
3. Copy the key (sk-ant-...)

Google (Gemini)

1. Sign in at https://aistudio.google.com/
2. Open Get API key → Create API key
3. Copy the key

Important: An API key grants access to your provider account and its billing. Store it securely and never expose it in code or logs.

---

2) Choose a Model

Enter a model ID that matches the selected provider. Some current examples:

Provider	Fast / cost-effective	High capability
OpenAI	gpt-4o-mini	gpt-4o
Anthropic (Claude)	claude-haiku-4-5	claude-opus-4-8
Google (Gemini)	gemini-2.0-flash	gemini-2.0-pro

Use a fast/cheap model for routine tasks (autocomplete, simple summaries) and a higher-capability model for complex reasoning. Model IDs change over time — check your provider's documentation for the latest available models.

---

3) Create the Service Account in MinuteView Console

1. Navigate to Console → Service Accounts
2. Click New
3. Set Service Account Type to AI Model
4. Fill in the fields:

Field	Description	Example
Name	A descriptive name (include provider/purpose)	AI-OpenAI-FastGeneral
Provider	The LLM vendor: OpenAI, Anthropic (Claude), or Google (Gemini)	OpenAI
Api Key	The API key for the selected provider	sk-proj-...
Default API Model	Model ID for the selected provider	gpt-4o-mini

5. Click Test Connection to verify (see note below)
6. Click Save

Test Connection behaviour

Live key validation currently runs for OpenAI accounts only. For Anthropic and Google accounts the key is not validated up front — if it is wrong, you will see the failure the first time an AI feature runs. Double-check the key and model ID when creating those accounts.

---

4) Recommended Strategy — Multiple Accounts

You can create several AI Model accounts and select the right one per feature:

• Fast/general — e.g. OpenAI gpt-4o-mini or Claude claude-haiku-4-5 for autocomplete, simple summaries, routine automation.
• Advanced — e.g. OpenAI gpt-4o or Claude claude-opus-4-8 for complex reasoning, cross-referencing, and high-accuracy analysis.
• Per-provider — keep separate accounts if you want to compare providers or isolate billing.

Assign accounts to features under Mesh Configuration → AI Settings, Global Settings → AI Assistant, and in individual AI workflow nodes.

---

5) Security Best Practices

• Store keys only in MinuteView's secure credential store — never in logs, URLs, or client-side code.
• Rotate keys regularly and after any suspected compromise.
• Set usage limits and billing alerts in your provider's dashboard.
• Use the minimum-capability model that meets each need to control cost.

---

6) Troubleshooting

Connection error / "Failed to reach provider"

• The most common cause is that the MinuteView server has no outbound internet access to the provider host. See Network Requirements above and whitelist the host for your selected provider.

Authentication failed / 401

• Verify the API key is correct and active, and that billing is set up with the provider.
• Confirm the key matches the selected Provider (an OpenAI key will not work on an Anthropic account).

Model not found / invalid model

• Check the model ID spelling and that your provider account has access to that model.
• Make sure the model belongs to the selected provider (e.g. don't enter a Claude model on an OpenAI account).

Anthropic/Google account "saved fine" but fails at runtime

• Test Connection does not live-validate these providers. Re-check the key and model ID, and confirm the host is reachable from the server.