Productive's AI runs on models from OpenAI, Anthropic, and Google. In the AI Assistant, and separately for each agent, you can choose which model powers your conversation and how much it reasons before answering, or stick with the defaults, which handle most tasks well.

📌 See AI: General Information to turn on Productive AI and manage your AI settings.

Your assistant's model and thinking effort apply to your conversations. Each of your assistant's triggers can also run on its own model.

You can switch models in two places:

Thinking effort sets how much the Assistant reasons before answering: Low, Medium, or High. Higher effort can take longer but produces more considered responses.

To change it, open Settings and choose a level under Thinking effort.

📌 Advanced models and higher thinking effort can draw more from your organization's monthly AI usage. Billing admins can track it in Settings > Usage.

👉 For more information, see How AI Usage Works.

Each trigger you set up for your assistant can run on its own model. Choose it from the instructions field in the trigger's Then step, where you write what your assistant should do.

This way, a quick daily check can run on a faster model, while a weekly project review can use a more advanced one.

👉 For more information, see Setting Up Triggers for AI Assistant.

Agents don't inherit your Assistant settings.
​
Each agent can run on a different model, at a different thinking effort. Set both on the agent's Info tab, under Agent details; they start on Auto. The choice applies to every run, in chat and on the agent's triggers.
​

👉 For more information on configuring an agent, Setting Up and Managing AI Agents.
​

In the model picker, models are grouped by provider: OpenAI, Anthropic, and Google. The default works well for most tasks, so change it only if you prefer something else.

📌 The specific models available may change over time.

📌 If your admin has restricted which models your organization can use, you'll only see the approved ones.

Admins can choose which AI models the whole organization can use in Productive, instead of leaving every model open to everyone.

This helps when you have an internal AI policy, want to standardize on the providers you already trust, or want to keep AI usage predictable. Only the models you allow appear in the model picker for everyone in your organization.

To set it up, go to Settings > AI:

Productive does not use your Customer Data or allow third-party providers to use it to train the machine learning models that power the Productive AI.

Your use of Productive AI does not grant Productive any rights or licenses to use your Customer Data to train machine learning models.