Dashboard

The Dashboard is Alfred’s main web experience. Use it when you want the full chat interface, file uploads, account settings, usage charts, model selection, module controls, and saved chat history.

Chat

The center of the Dashboard is the chat. Type a message in the chat bar and press Enter to send. Use Shift+Enter for a new line.

Alfred streams responses as they arrive. While he works, you may see status updates such as “Searching the web”, “Reading website”, “Creating reminder”, “Running code”, or “Uploading image”. These updates show what Alfred is doing without exposing the internal machinery.

Sometimes Alfred starts a turn himself, because a background task came back or a job reported in. Rather than a message you never typed, the chat shows a quiet line saying his runtime reported back, which you can open to read what it actually said. A chat follows those turns as they happen, so you don’t need to refresh to see one arrive.

Uploads

Use the plus button in the chat bar to upload files. You can also drag and drop files into the Dashboard or paste files when the chat area is focused.

Alfred gains direct access to most text-based files, and can access other file uploads through some of his tools. For example, Alfred may not be able to interact with a .zip file directly but can use it in his sandbox tools to interact with it.

Large text files become Cards, which is how Alfred works with them without dumping the whole thing into the conversation.

Panels

There are two panels on the dashboard, a left panel for conversation-wide controls and a right panel for account settings.

Chat Panel (Left)

On the left is the Chat Panel. One important piece of information you can get from here is your usage information. You can use View Usage in the left panel to see how much Alfred usage you have used over specific periods. See Usage for the full explanation.

You’ll also see your conversation history in the left panel. Note that conversations between Alfred and his agents are also visible here so you can see Alfred’s internal work process with his agents.

Below View Usage is Jobs, which opens the jobs popup. The button carries a count of the jobs running right now plus anything Alfred is waiting on you for, so it’s the one thing to look at when you’re wondering what he’s up to.

Inside the popup:

• What needs you sits at the top: open questions and approvals, each with a box to answer in and Approve and Deny buttons where the job asked permission.
• Then the job list, filtered by All, Active, Waiting or Finished, with each job’s status, the phase it’s on, what it has spent today against your allowance, and its sub-jobs opened up underneath it.
• Opening a job shows its goal, its plan as a checklist, and its journal as a timeline grouped by shift, each entry saying who wrote it, what that shift was there to do, and which model it ran on. Show older pages back through a long journal.
• The controls are at the bottom: autonomy and how long it holds for, the always-ask list, and Stop.

See Jobs for what all of that means.

Settings Panel (Right)

On the right is the Settings Panel. It contains account and preference controls. You can manage your account and profile from here along with your preferred AI model, installed modules, and performance settings.

From here you can also set your preferred model. The chosen model affects quality, speed, and usage cost. If your usage is high, Alfred may temporarily switch to a cheaper model to preserve your allowance. See Models for more information on AI models.

Use Modules to see Alfred’s available abilities. Modules can be installed, uninstalled, enabled, or disabled depending on what the module supports and what your plan allows. See Modules.

The Dashboard includes performance controls that trade off cost, context, and depth:

• Minimal Reasoning: This minimizes Alfred’s allowed reasoning depth, which can save costs and speed up responses, but may result in lower quality responses.
• Memory Compression: This compresses all previous individual messages, tool calls and tool results. This lowers the context Alfred has to carry over time while still keeping the overall concepts and shape of the history.
• Limit Reason Memory: This prunes Alfred’s memory of his own reasoning steps, which can save costs and speed up responses.

These are the first thing to reach for if you’re burning through allowance faster than you’d like. Memory Compression in particular pays off on long conversations. See Usage.

Job defaults live here too: how much your jobs may spend in a day, as a cash amount or a percentage of your usage limit, how freely a new job may act, and what every job must ask you about regardless. These are the defaults new jobs start with, and each job keeps its own copy you can change from the jobs popup.

Keyboard and layout

Enter sends, Shift+Enter starts a new line. On smaller screens both panels collapse behind panel buttons in the chat header.

Common fixes

If a model you expect isn’t in the picker, check your plan on Patreon and confirm your Patreon account is connected.

If Alfred won’t use an ability you expect him to have, check that its module is enabled. See Modules.

If answers feel worse than usual in a long conversation, start a new one. Long histories crowd out the thing you’re actually asking about.