How SafeAppeals Uses AI

The AI assistant helps you research, draft, and organize case work. It is a drafting aid—not legal advice and not a court-ready filing.

Data Flow and Confidentiality

Your case files stay on this computer. When you ask the AI assistant a question, the text you send — and any documents you attach — go to SafeAppeals Cloud to generate the answer, then to the AI model provider. Your prompts and documents are used only to generate your answer and are handled under the model provider's retention policy. You remain responsible for reviewing everything the assistant produces: its output is a drafting aid, not legal advice and not a court-ready filing. Read our Privacy Policy for details.

Client-Consent Guidance for Your Practice

If you represent clients, consider whether your engagement letters, privacy notices, and local ethics rules (including ABA Formal Opinion 512 and Florida Bar Ethics Opinion 24-1 guidance) require informing clients when AI tools process confidential information. SafeAppeals keeps case files on the local machine by default; only the prompts and attachments you send in chat leave the device for model inference.

• Disclose AI use when your jurisdiction or engagement terms require it
• Do not paste sealed, privileged, or restricted material unless your client engagement allows it
• Review every citation and factual claim before relying on assistant output
• Prefer approve-before-write settings when working in live case workspaces

Basic Usage

Opening the Chat

Access the AI assistant in several ways:

• Click the Chat icon in the left activity bar
• Press Ctrl+L to focus the chat input
• Press Ctrl+Shift+L to open the chat with your current selection

Sending Messages

• Type your message in the input field
• Press Enter to send (or Shift+Enter for a new line)
• Wait for the AI to respond—longer responses may take a few seconds

Adding Context

Help the AI understand your needs by providing context:

• @-mention files — Type @ and select files to include in your message
• Select text first — Select text in the editor, then press Ctrl+Shift+L
• Describe your case — The more context you provide, the better the response

Modes

SafeAppeals offers modes tailored for legal work (built-in Ask and Agent, plus custom Plan):

Changing Modes

1. Look for the mode dropdown at the top of the chat sidebar
2. Click to open the dropdown
3. Select Ask, Agent, or Plan

For the full Plan Mode workflow (CreatePlan, reviewPlan, plan file locations, and legal gitignore guidance), see the Plan Mode guide.

Agent Tools

In Agent mode, the assistant invokes built-in tools from SafeAppeals extensions — email search and drafts, billing timer, Word/Excel editing, timeline events, Private Search, and more. Sensitive actions often ask for your confirmation first.

• Email — Read, organize, tag, link to cases, and create drafts. The agent cannot send; you send from the Email Dashboard.
• Documents — Create and edit .docx and .xlsx files, open in the visual editor or headlessly.
• Timeline & timer — Add events, query deadlines, start/stop the legal billing timer.

See the full catalog in the Agent Tools guide.

Private Search vs Cloud Chat

Private Search is separate from cloud chat. It indexes case files and core_references/ on this computer so you can find passages without uploading whole files for every search. Setup and optional model installs ask for consent first. Searching the private index does not spend credits and does not require the AI to run.

When you use Ask or Agent mode, the assistant can call Private Search tools (reference scope, case-file scope, or both) after the index is set up. Cloud chat still only sends the text and attachments you choose when you ask a question. See Core References & Private Search.

Quick Edit (Ctrl+K)

Quick Edit is a powerful feature for making inline changes to your documents:

1. Select the text you want to modify in the editor
2. Press Ctrl+K
3. Describe the change you want (e.g., "make this more formal")
4. Press Enter to apply
5. Review the changes and accept or reject them

Quick Edit Ideas

Best Practices for Legal Research

Effective Prompting

• Be specific — Include jurisdiction, type of claim, and specific issues
• Provide context — Share relevant case details when asking for advice
• Ask for sources — Request citations and ask the AI to verify them
• Iterate — Follow up with clarifying questions

Verification Checklist

Token Usage

Each message uses tokens from your credit balance. Token usage depends on:

• Input tokens — Your message length plus any files/context you include
• Output tokens — The length of the AI's response
• Model choice — Different models have different token costs

View your current balance in the Dashboard.

Next Steps