Skip to main content
Support Us
Docs / / AI Features in AcaMail

AI Features in AcaMail

Comprehensive guide to AcaMail's local AI capabilities including email triage, semantic search, and Reviewer 2 mode.

AI Features in AcaMail

AcaMail brings powerful AI capabilities to academic productivity—all running 100% locally on your machine. This guide covers each AI feature and how to use it effectively.

Privacy-First AI

Before diving into features, understand what makes AcaMail’s AI different:

  • Local Processing: All AI runs on your GPU via Ollama
  • No Cloud AI: Zero requests to OpenAI, Anthropic, or any external AI service
  • Your Data Stays Yours: Emails, documents, and conversations never leave your machine

Email Intelligence

AI Triage

The triage system automatically analyzes incoming emails and assigns:

Categories:

  • Urgent: Requires immediate action (deadlines, critical requests)
  • FYI: Informational, no action needed
  • Academic: Research, grants, publications, teaching
  • Administrative: HR, compliance, institutional
  • Meetings: Invites, scheduling, agendas
  • Personal: Non-work correspondence
  • Newsletters: Mailing lists, digests

Priority Scores (1-5):

  • 5: Critical, time-sensitive
  • 4: Important, respond today
  • 3: Normal priority
  • 2: Low priority, batch process
  • 1: Optional, can be deferred

How Triage Works

  1. Click “Triage Inbox” or press Ctrl+T
  2. AI analyzes unread emails based on:
    • Content keywords and urgency markers
    • Sender importance (learned from your behavior)
    • Your academic role
    • Deadline proximity
  3. Results appear as badges on each email

Traditional search finds exact keyword matches. Semantic search finds emails by meaning:

Example Queries:

  • “emails about the NSF grant deadline” — finds all related correspondence
  • “discussions about methodology concerns” — surfaces critique threads
  • “student asking for deadline extension” — finds patterns across students

How to Use:

  1. Press Ctrl+K to open search
  2. Type your query naturally
  3. Results are ranked by semantic similarity

Smart Reply

AcaMail can draft replies based on email context:

  1. Open an email requiring response
  2. Click “Smart Reply” or press R
  3. AI generates 2-3 suggested responses
  4. Select one, edit as needed, and send

The AI learns your writing style over time, making suggestions more personalized.

Daily Digest

Each morning, AcaMail generates a summary of:

  • High-priority items from the past 24 hours
  • Upcoming deadlines mentioned in emails
  • Calendar events requiring preparation
  • Suggested focus areas for the day

Access via Home > Daily Digest or enable morning notifications.

Document Analysis

Summarization

Open any document from Drive and click “Summarize”:

  • Generates a concise overview
  • Extracts key points and arguments
  • Identifies sections needing attention

Reviewer 2 Mode

The signature AcaMail feature for academic writing:

What It Does: Provides harsh, constructive criticism mimicking the infamous “Reviewer 2” from academic peer review.

How to Use:

  1. Open or paste your document text
  2. Enable “Reviewer 2 Mode” toggle
  3. Request analysis with prompts like:
    • “Critique this abstract”
    • “Find weaknesses in my methodology”
    • “What would Reviewer 2 say about this?”

Example Output:

“Your broader impacts section is vague and unconvincing. The claim that this work will ‘advance the field’ is meaningless without concrete outcomes. Additionally, the experimental design lacks a clear control condition, which any competent reviewer will immediately flag.”

Spreadsheet Analysis

For Google Sheets:

  • Data Insights: AI identifies patterns and anomalies
  • Formula Suggestions: Get help with complex calculations
  • Summary Statistics: Quick overview of your data

Presentation Review

For Google Slides:

  • Speaker Notes Generation: AI creates notes for each slide
  • Flow Analysis: Identifies narrative gaps
  • Audience Targeting: Suggestions for different audience types

Calendar Intelligence

Meeting Briefs

Before any meeting, generate a preparation document:

  1. Open the calendar event
  2. Click “Generate Brief”
  3. AI creates a document with:
    • Attendee backgrounds and roles
    • Recent email history with each participant
    • Relevant documents from Drive
    • Suggested agenda items
    • Prior meeting outcomes

Scheduling Suggestions

When composing new events, AI suggests:

  • Optimal times based on your patterns
  • Duration estimates based on topic complexity
  • Attendees who should be included

Voice Features

Local Speech-to-Text

AcaMail uses Whisper large-v3-turbo running on your GPU:

  • Activation: Click microphone or press V
  • Continuous Mode: Keep talking, AI transcribes in real-time
  • Interrupt: Speaking stops AI response generation

Voice Commands

While voice is active, use natural commands:

  • “Search for emails about [topic]”
  • “Schedule a meeting with [person]”
  • “Summarize this document”
  • “Draft a reply”

Learning Engine

AcaMail continuously learns from your behavior:

What It Tracks

  • Email Actions: Opens, replies, archives, deletes
  • Response Patterns: How quickly you respond to different senders
  • Search Behavior: Topics you frequently search for
  • AI Feedback: Which suggestions you accept or reject

How It Adapts

  • Sender Importance: VIPs are identified from your response patterns
  • Category Preferences: Your custom priority rules emerge over time
  • Role Detection: Academic role is inferred and adjusted
  • Writing Style: Smart replies match your voice

Viewing Learning Insights

Navigate to Settings > Learning Dashboard to see:

  • Detected sender importance scores
  • Category priority adjustments
  • Behavioral patterns identified
  • Override any learned preferences

Chain-of-Thought Reasoning

AcaMail uses DeepSeek-R1, a model with native chain-of-thought capabilities:

Visible Reasoning

When AI processes complex requests, you can see its thinking:

  • Click the “Show Reasoning” dropdown
  • View the <think> block showing decision process
  • Understand why AI reached its conclusions

Benefits for Academics

  • Transparent Analysis: See how AI evaluated your writing
  • Debugging: Understand unexpected suggestions
  • Teaching: Use reasoning examples in methodology courses

Tips for Effective Use

Writing Better Prompts

  • Be Specific: “Critique the methodology section” > “Review this”
  • Provide Context: “As a reviewer for Nature…” helps target feedback
  • Ask Follow-ups: Drill into specific concerns

Optimizing Performance

  • Close Other GPU Apps: Games, video editors compete for VRAM
  • Batch Operations: Run triage once daily, not per-email
  • Use Keyboard Shortcuts: Ctrl+K (search), Ctrl+T (triage), R (reply)

Privacy Considerations

Even with local AI, practice good data hygiene:

  • Don’t paste confidential third-party data unnecessarily
  • Review Smart Reply suggestions before sending
  • Understand that learning data persists locally

For more AI tips, visit our community forum where academics share workflows and prompts.