Connecting an AI Provider, or Running One Yourself

Every AI feature runs on a provider you choose and pay for. There is no shared pool and no Pental credit. The screen is Settings, then AI, and it holds the provider, the key, the model and a couple of limits.

What you can point it at

• Ollama, local or self-hosted, which the screen marks as recommended. Nothing leaves the machine.
• Groq, which is fast and has a free tier, so it is a reasonable place to try the features before committing.
• OpenAI.
• Anthropic Claude.
• A custom endpoint, declared as either OpenAI-compatible or Anthropic-compatible, for anything you host yourself or buy elsewhere.

For a hosted provider you paste a key and pick a model. The key is encrypted on write into your own database, alongside your findings and under the same policies. For Ollama or a custom endpoint you give a base URL instead, and the screen discovers the models available and lets you choose from what it found rather than making you type a name.

The two settings worth understanding

• Maximum tokens per request. Two thousand is a sensible default for most self-hosted models; hosted models can take more.
• Keeping a large model resident on the GPU, which is worth switching on for anything from about fourteen billion parameters up, because the reload cost otherwise dominates.

Local models have no rate limits, so the platform can take as long as a job needs. Hosted providers do, so the same work is paced to stay inside them. Neither runs a reduced-quality pass to save time.

The prompts are yours

Every field the platform can write has a prompt you can read and edit, in the same settings area as the field itself. That covers the built-in fields and each of your own custom ones. Nothing is hidden inside the product, so if the output is not in your house voice, the fix is a prompt you can see rather than a support ticket.