Provider Configuration

Provider Configuration

Meetily supports 8 AI providers for generating meeting summaries. Each provider has different strengths -Ollama runs locally for privacy, Claude and OpenAI offer powerful cloud models, Groq delivers ultra-fast inference, OpenRouter gives access to models from multiple providers, and Custom connects to any OpenAI-compatible server you control. This page covers summary model providers. For transcription engine settings, see Retranscription.

Think of it like setting up email accounts -you add a provider, enter credentials, and pick a model. Once configured, Meetily uses that provider whenever you generate a summary.

Configure providers in Meeting Details > Preferences > Summary Model.

Quick Start

1. Open any meeting from the sidebar
2. Click the Preferences icon (gear) in the top-right toolbar
3. Select the Summary Model tab
4. Browse the provider cards -click one to select it
5. Click the Settings icon (gear) on the provider card to configure it
6. Enter your API key, endpoint URL, or model selection
7. Click Save -you're ready to generate summaries

Want to use a different AI model?
  → Open Preferences > Summary Model
  → Click a provider card
  → Click Settings (gear icon)
  → Enter credentials and pick a model
  → Save -done!

Available Providers

Provider Cards

When you open the Summary Model tab, you see a grid of provider cards:

Click any card to select that provider. The selected card shows a checkmark and is highlighted in blue.

Configuring Each Provider

Built-in AI (Local)

Runs a small AI model directly inside Meetily. No setup required.

Setup:

1. Click the Built-in AI provider card
2. That's it -no configuration needed

The configuration modal shows an info box: "Offline AI runs locally, no API key required."

Best for quick summaries when you don't have Ollama or cloud access set up.

Hosted AI

A pre-configured cloud service provided by Meetily.

Setup:

1. Click the Hosted AI provider card
2. No configuration needed -it's pre-configured

Claude (Cloud)

Uses Anthropic's Claude models for high-quality, nuanced meeting summaries.

Setup:

1. Click the Claude provider card
2. Click Settings (gear)
3. Enter your API Key in the password field:
   • The field shows dots by default (like a password)
   • Click the eye icon to reveal the key
   • Click again to hide it
4. Click Save

Where to get an API key: Sign up at console.anthropic.com(opens in new tab) and create a key in the API Keys section.

OpenAI (Cloud)

Uses OpenAI's API for cloud-based summarization with models like GPT-4.

Setup:

1. Click the OpenAI provider card
2. Click Settings (gear)
3. Enter your API Key in the password field:
   • The field shows dots by default (like a password)
   • Click the eye icon to reveal the key
   • Click again to hide it
4. Click Save

Where to get an API key: Sign up at platform.openai.com(opens in new tab) and create a key in the API Keys section.

Groq (Cloud)

Ultra-fast inference powered by Groq's LPU hardware. Great for real-time summaries.

Setup:

1. Click the Groq provider card
2. Click Settings (gear)
3. Enter your API Key in the password field
4. Click Save

Where to get an API key: Sign up at console.groq.com(opens in new tab) and create a key.

Ollama (Local)

Ollama runs large language models locally on your machine. No internet needed, no API key required.

Setup:

1. Click the Ollama provider card to select it
2. Click the Settings (gear) icon on the card
3. The configuration modal opens with these fields:

4. If the endpoint is correct, click Refresh next to "Select Model"
5. A spinner appears while loading models from your Ollama server
6. Select a model from the dropdown (e.g., llama3.1, mistral)
7. Click Save

OpenRouter (Cloud)

A single API key to access models from OpenAI, Anthropic, Meta, and more. Great for flexibility.

Setup:

1. Click the OpenRouter provider card
2. Click Settings (gear)
3. Enter your API Key in the password field
4. Click Save

Where to get an API key: Sign up at openrouter.ai(opens in new tab) and create a key in your account settings.

Custom (OpenAI-Compatible)

Connect to any server that implements the OpenAI Chat Completions API -LM Studio, vLLM, text-generation-webui, or your own deployment.

Setup:

1. Click the Custom provider card
2. Click Settings (gear)
3. Configure these fields:

4. Click Save

The Configuration Modal

When you click the Settings icon on a provider card, a configuration modal opens.

Modal Layout

API Key Field Behavior

For providers that need an API key (Claude, OpenAI, Groq, OpenRouter, Custom):

• Hidden by default -shows dots like a password field
• Eye icon -click to toggle visibility (show/hide the key)
• Autocomplete disabled -prevents browsers from auto-filling
• Placeholder text -shows sk-... for OpenAI

Model Dropdown (Ollama)

For Ollama, the model dropdown works differently from other providers:

1. The dropdown is initially empty or shows previously saved model
2. Click Refresh to fetch available models from your Ollama server
3. A spinner appears while loading: "Loading models..."
4. Models appear in a scrollable dropdown list
5. Select your preferred model
6. The selection is saved when you click Save

If the Ollama server is unreachable, the refresh shows an error state.

Switching Providers

To switch your active provider:

1. Open Preferences > Summary Model
2. Click a different provider card
3. The new provider is selected immediately (checkmark appears)
4. If not yet configured, click Settings to set it up
5. The next summary generation uses the new provider

Each provider's configuration is saved independently. Switching away from a provider and back restores all its settings -API keys, endpoints, and model selections are preserved.

How It Works

When you generate a summary, Meetily routes the request to your selected provider:

Transcript → Selected Provider → LLM Model → Summary
                  ↓
     ┌────────────┼────────────────────────────┐
     ↓            ↓                            ↓
  Local        Cloud APIs                  Custom
  (Built-in,   (Claude, OpenAI, Groq,      (any OpenAI-
   Ollama)      OpenRouter, Hosted)         compatible)

1. Transcript is prepared: The full meeting transcript is formatted as a prompt
2. Template applied: Your selected summary template defines the output structure
3. Provider called: The request goes to your chosen provider's API endpoint
4. Response parsed: The LLM's response is parsed and displayed as the AI summary
5. Saved locally: The summary is stored in your local database

Real-World Examples

Example 1: Setting up Ollama for fully local AI

"I want everything to run on my machine -no cloud, no internet, complete privacy."

Setup: Install Ollama → pull a model (ollama pull llama3.1) → open Meetily Preferences > Summary Model → select Ollama → Settings → verify endpoint is http://localhost:11434 → click Refresh → select llama3.1 → Save. All summaries now run locally.

Example 2: Adding OpenAI for higher quality summaries

"I want to use GPT-4 for the best possible meeting summaries."

Setup: Go to platform.openai.com and copy your API key → open Meetily Preferences > Summary Model → select OpenAI → Settings → paste your API key → Save. The next summary generation uses OpenAI's models.

Example 3: Connecting to LM Studio on your network

"I run LM Studio on a separate workstation with a powerful GPU. I want Meetily on my laptop to use it."

Setup: Start LM Studio's server on the workstation → open Meetily Preferences > Summary Model → select Custom → Settings → endpoint: http://192.168.1.50:1234/v1 → model: llama-3.1-8b-instruct → API key: leave empty → Save. Meetily sends transcripts to your workstation for summarization.

Example 4: Quick summaries with Built-in AI

"I'm in a coffee shop with no internet and Ollama isn't installed. I just want a basic summary."

Setup: Open Preferences > Summary Model → select Built-in AI → no setup needed → generate a summary. The built-in model produces a basic summary instantly, right on your device.

Screenshots Guide

These are the key screens worth capturing for visual reference: