Chat Controls

Pentos gives you fine-grained control over what context, tools, and AI behavior apply to every message you send. The chat input menu houses several controls that you can change at any time. Each one carries over from your previous message, so you only need to change them when you want something different.

Where to Find Them

Open the chat input menu by clicking the + button in the bottom left corner of the chat input area. From there, you can toggle any of the controls below for your next message.

The full list of controls, in the order they appear:

  • Personalize to share your profile with the AI.
  • Location to share your approximate region.
  • Web Search to let the AI pull live information.
  • Use My Key to route through your own API key.
  • Context modes (Lean, Standard, Max) to control how much chat history the AI receives.
  • Effort to set how much the AI reasons before it answers.
  • Connectors to give the AI access to your connected apps.

Context Modes

The Context control sets how much of your conversation the AI receives: Lean (your last 10 messages, the energy-saving default), Standard (as much recent history as fits the model’s standard window), or Max (the model’s full context window, available on paid plans). We recommend leaving it on Lean by default and pairing it with the Pentos Smart Router, which keeps both the model and the context small. See Energy Usage for how the savings are estimated.

For the full breakdown, see Context Modes.

Effort

The Effort control sets how much the AI reasons before it answers: Auto, None, Low, Medium, High, or Max (available on paid plans). Auto lets Pentos pick a sensible effort for the model. Not every model offers every level, and a level a model doesn’t offer runs at the nearest one it does. Choosing High adds 1 point to each response and Max adds 3; Auto never adds points. Effort also steers the Smart Router.

Personalize

When enabled, the AI will receive additional information about you, including your name, occupation, interests, and any additional information you have in your Pentos Profile, along with the memories you have saved in Memory. This allows the AI to tailor its responses to you based on your background and the durable facts Pentos has learned about you over time.

For example, if you are an engineer, the AI may respond with more technical data and granular information, whereas if you are a student, it may respond with additional tips, tricks, and resources to help you learn.

Personalization data is sourced from your Pentos Profile and your Memory, and nothing more. To update your profile, head to your Pentos Profile. To view, add, edit, or delete what Pentos remembers about you, head to Settings > Data > Memories.

Location

When enabled, Pentos will determine your approximate location using your IP address and pass it along to the AI. This is useful for questions about local time, regional information, or nearby resources.

To help protect your privacy, Pentos determines your location using a private table in our database that routes IP ranges to specific locations. For example, an IP address within the range of 32.123.49.0/24 is tied to New York City, so we would tell the AI that you are currently in New York City.

If we are unable to determine your current location, Pentos will use the default location you set in your Pentos Profile instead. This way, we don’t rely on, or share your data with, a third-party geolocation service.

NOTE: Since we use your IP address to determine your current location, the use of VPNs and other technologies may render inaccurate results. Additionally, Pentos does not pass along IP addresses to the AI, only estimated geolocations (City, State, Country).

When enabled, the AI can search the web for current information before responding. This is useful for questions about recent events, live data, or anything beyond the model’s training data. While Web Search is on, each response costs 1 more point, whether or not the AI searches. We recommend leaving this feature on by default for the most up-to-date information. However, for users who don’t want the AI to conduct web searches using the information from the message, this can be turned off. This is particularly useful for certain compliance requirements.

Use My Key (BYOK)

When enabled, your messages for the selected provider are routed through your own API key instead of Pentos’s connection. Messages sent with your own key never cost points. Images the AI creates or edits in that message use your key too when you’ve added one for the image model’s provider; otherwise they use Pentos’s connection and cost points as usual. This is useful when you want better control and insight into your AI usage, or want to use Pentos on a free plan while still accessing top models.

If you enable the toggle but haven’t added a key for the current provider yet, Pentos will prompt you to add one. Claude Fable 5.1 and GPT 6 Astra always run on your own key, so the toggle is locked on while one is selected. Note that BYOK is not supported with the Pentos Smart Router. To use your own keys, select a specific model from the model selector first.

For the full setup walkthrough and key security details, see Bring Your Own Key.

Connectors

When enabled, the AI can use your data and take actions in connected apps like Gmail, GitHub, Slack, and over a thousand others, directly from your chat. The AI automatically determines when a Connector is needed for your message, no special syntax required.

Connectors are off-limits to the AI unless you have Connectors enabled for that message and have authorized the relevant app. You can connect or disconnect apps at any time from the Manage button next to the toggle. Connectors need a paid plan: on Free, turning them on opens the upgrade options.

For the full app catalog, OAuth flow, and privacy details, see Connectors.

Updating Your Defaults

Each control has a default that new chats start with. To change your defaults, go to Settings > Chats > Preferences and update AI Settings: the toggles, plus the Context and Effort dropdowns. Changes save right away. Effort starts at your default, in chats and in the AI Sidebar, and a level you pick in the chat input menu stays until you change it or reload Pentos.

Read Aloud Voice

When you press Read on a message, Pentos reads it aloud in your chosen voice. To pick a voice, go to Settings > Chats > Preferences and click Voice under AI Preferences. Press play in the corner of any voice to hear a short sample, then select the one you like. Your choice saves right away and applies to Read in both chats and Collab.

  • Default: Eve.
  • While it reads: a small player appears at the top of the screen. Once the voice starts, press Stop to stop reading. The audio still finishes loading, so reading that message again later starts right away.
  • Length limit: messages up to 60,000 characters can be read aloud. For longer messages, Pentos lets you know instead.

Context Disclaimer

Please note that if a control is enabled for a message and then disabled for a subsequent message, the AI may still see related data from earlier in the conversation. For example, if you have Location enabled to ask the AI for the weekend weather, it will respond with the weather for your location. If you then disable Location on your next message, the AI will still see your location from the last message as part of the chat history, even though the toggle is off.

Approving ActionsFormatting