Skip to main content
Freiya’s AI chat assistant is your intelligent research companion. Ask questions, explore topics, analyze papers, and get instant insights powered by state-of-the-art language models.
Full chat interface showing a conversation about research methodology, with the model selector dropd

​
Overview

The AI chat system uses LangChain to orchestrate conversations with multiple AI providers including OpenAI and Google AI. Your conversations are saved automatically, searchable, and can be organized into collections or shared with colleagues. Key capabilities:
  • Multiple AI models for different use cases
  • Context-aware conversations with memory
  • Real-time streaming responses
  • Usage tracking and limits by tier
  • Save, organize, and share conversations

​
Available AI models

Freiya provides access to different AI models based on your subscription tier. Each model is optimized for specific tasks and use cases.

​
Model types

Best for: Everyday research questions, quick lookups, general inquiriesStandard models provide solid performance for most research tasks. They’re fast, reliable, and included in all tiers including Free.Use cases:
  • Explaining basic concepts
  • Summarizing paper abstracts
  • Generating research questions
  • Brainstorming ideas
  • Quick fact checking
Your subscription tier determines which models you can access. See subscription tiers for model availability by plan.

​
Starting a conversation

Creating and managing chats in Freiya is simple and intuitive.

​
Create a new chat

1

Open chat

Click Chat in the sidebar or press Ctrl/Cmd + N
2

Select a model

Choose your preferred AI model from the dropdown at the top
3

Type your message

Enter your question or prompt in the input box
4

Send

Press Enter or click the send button (use Shift + Enter for line breaks)
Close-up of the chat input area showing the message input box with placeholder text, send button, an
Start with a clear, specific question. The more context you provide, the better the AI can assist you.

​
Example prompts

Good prompts: 
Explain the difference between supervised and unsupervised learning
in the context of medical image analysis.

Summarize the key findings from the paper "Attention Is All You Need"
and explain why the transformer architecture was revolutionary.

What are the main limitations of this methodology, and how might
I address them in my research design?
Less effective prompts: 
Tell me about machine learning
What's this paper about?
Help with research

​
Chat interface features

​
Message display

All messages support rich formatting:
  • Markdown rendering - Bold, italics, lists, and more
  • Code blocks - Syntax-highlighted code in 100+ languages
  • Math equations - LaTeX math rendered with KaTeX
  • Tables - Formatted tables for structured data
  • Links - Clickable URLs and references
Chat message showing various formatted elements: a code block with syntax highlighting, a mathematic

​
Message actions

Hover over any message to access quick actions:
ActionDescriptionShortcut
CopyCopy message content to clipboardCtrl/Cmd + C
RegenerateGenerate a new responseCtrl/Cmd + R
EditEdit your message and resendCtrl/Cmd + ↑
DeleteRemove message from chatDelete
VoteThumbs up/down for feedback-

​
Streaming responses

Responses stream in real-time, so you see the AI “thinking” as it generates text. You can:
  • Read as it types - Start reading before the response finishes
  • Stop generation - Click stop if you have enough information
  • Copy partial - Copy text even while streaming

​
Usage tracking

Freiya tracks your AI usage to ensure fair access and manage costs. Each model type has separate limits.

​
How usage works

1

Monthly allocation

Your tier determines how many messages you get per model type each month
2

Per-message deduction

Each message you send consumes one message credit for that model
3

Real-time tracking

See your remaining messages for each model in the interface
4

Monthly reset

All limits reset on the 1st of each month
Usage indicator showing message count like '42/100 messages remaining this month' with a small progr

​
View usage statistics

Navigate to Settings → Usage to see:
  • Messages used per model type
  • Messages remaining
  • Next reset date
  • Historical usage charts

​
Managing conversations

​
Chat titles

Chats are automatically titled based on your first message. You can rename them anytime:
  1. Click the chat title or triple-dot menu
  2. Select Rename
  3. Enter a new title
  4. Press Enter to save
Chat title being edited inline, showing the editable text field with a cursor

​
Pinning chats

Pin important conversations to keep them at the top of your sidebar:
  1. Hover over a chat in the sidebar
  2. Click the pin icon
  3. Pinned chats appear in a “Pinned” section at the top
Use cases for pinning:
  • Active research projects
  • Frequently referenced conversations
  • Important methodology discussions
  • Ongoing brainstorming sessions

​
Chat organization

Sidebar features:
  • Search - Find chats by title or content
  • Filter by date - Today, Yesterday, Last 7 days, etc.
  • Filter by tags - Organize with custom tags (coming soon)
  • Sort options - By date, title, or relevance
Chat sidebar showing pinned chats at the top, followed by date-grouped chats (Today, Yesterday, Last

​
Deleting chats

Remove conversations you no longer need:
  1. Open the chat or hover in sidebar
  2. Click the triple-dot menu
  3. Select Delete
  4. Confirm deletion
Deleting a chat is permanent and cannot be undone. Consider exporting important chats before deleting.

​
Chat collections

Group related conversations into collections for better organization.

​
Creating chat collections

1

Navigate to Collections

Click Collections in the sidebar
2

Create new collection

Click New Chat Collection
3

Name and configure

Enter name, description, and choose a color
4

Add chats

Add existing chats to your new collection
Example collections:
  • “Thesis Research” - All chats related to your thesis
  • “Methodology Discussions” - Chats about research methods
  • “Literature Review” - Conversations about papers you’ve read
  • “Project Ideas” - Brainstorming and planning chats
Free tier: 2 chat collections. Pro/Nova: 20 collections.
Learn more about collections →

​
Sharing conversations

Share interesting conversations with colleagues, on social media, or for reference.
1

Open the chat

Navigate to the chat you want to share
2

Click Share

Click the share icon or Share button
3

Generate link

Freiya creates a unique, read-only URL
4

Copy and share

Copy the link and share it anywhere
Share dialog modal showing a generated shareable link with a 'Copy Link' button and view count displ

​
Share features

  • Read-only access - Recipients can view but not edit
  • No login required - Anyone with the link can access
  • View tracking - See how many times it’s been viewed
  • Deactivate anytime - Revoke access whenever you want
  • Access token security - Unique token prevents guessing
Shared chats are publicly accessible to anyone with the link. Don’t share sensitive research before publication.

​
Paper-linked chats

Create chats specifically about a paper to get context-aware responses. Method 1: From paper detail page
  1. Open a paper in your library
  2. Click Chat about this paper
  3. New chat opens with paper context
Method 2: During chat
  1. Mention the paper title in your message
  2. Freiya suggests linking the paper
  3. Confirm to add context
Benefits of paper-linked chats:
  • AI has full paper metadata
  • More accurate, contextual responses
  • References specific sections
  • Links preserved in chat
Chat interface showing a paper-linked conversation with a small paper card/badge at the top indicati

​
Context and memory

Freiya maintains conversation context throughout your chat session.

​
How context works

  • Full chat history - AI sees all previous messages
  • Smart trimming - Long chats are condensed intelligently
  • Persistent memory - Context saved between sessions
  • Token management - Automatic optimization for long conversations
For best results, keep conversations focused on a single topic. Start a new chat for different subjects.

​
Advanced features

​
Model comparison

Compare responses from different models:
  1. Get a response from one model
  2. Switch to another model
  3. Ask the same question
  4. Compare the different approaches

​
Export conversations

Export chats for reference or archival:
  1. Open the chat
  2. Click Export
  3. Choose format (Markdown, PDF, or Text)
  4. Download the file
Use cases:
  • Archive important conversations
  • Include in research notes
  • Share with collaborators
  • Document AI assistance in your work

​
Chat with documents (coming soon)

Upload documents directly to chats for AI analysis:
  • Ask questions about uploaded PDFs
  • Analyze data files
  • Extract information from documents
  • Compare multiple documents

​
Best practices

Be specific

Provide context and details in your questions for better responses

Iterate

Refine your questions based on responses. Ask follow-ups for clarity

Verify information

Always verify AI responses against source materials

Organize chats

Use collections and clear titles to keep conversations organized

​
Effective prompting tips

  1. Start with context - Explain what you’re working on
  2. Be specific - Ask precise questions rather than broad ones
  3. Request format - Specify if you want lists, paragraphs, or tables
  4. Provide examples - Show the AI what you’re looking for
  5. Follow up - Build on previous responses with clarifying questions
Example progressive conversation: 
You: I'm researching neural network optimization for medical imaging.

You: What are the current state-of-the-art optimization algorithms
used in this field?

AI: [Provides response about Adam, RMSprop, etc.]

You: Can you compare Adam and RMSprop specifically for small
medical imaging datasets (< 1000 samples)?

AI: [Provides detailed comparison]

You: What would be the best approach for my dataset of 500
chest X-rays?

​
Troubleshooting

You have three options:
  1. Wait until the 1st of next month for your reset
  2. Switch to a different model type
  3. Upgrade your tier for higher limits
AI models can make mistakes. Always:
  • Verify against source materials
  • Cross-check important facts
  • Ask for clarification or sources
  • Use multiple models for important queries
Chats are saved automatically, but:
  • Check if you’re logged into the correct account
  • Look in the “All Chats” view, not just “Recent”
  • Check if you accidentally deleted it
  • Contact support if the issue persists
This can happen due to:
  • Token limits on the model
  • Network interruption
  • Click “Regenerate” to try again
  • Or ask the AI to continue with “Please continue”
You can use AI as a research assistant, but:
  • Always write in your own words
  • Use AI for brainstorming and understanding
  • Cite sources properly, not the AI
  • Check your institution’s AI usage policies
  • Never submit AI-generated text as your own work

​
Keyboard shortcuts

Speed up your chat workflow:
ActionShortcut
New chatCtrl/Cmd + N
Send messageEnter
New lineShift + Enter
Edit last messageCtrl/Cmd + ↑
RegenerateCtrl/Cmd + R
Copy messageCtrl/Cmd + C (hover on message)
Delete messageDelete (hover on message)
See all shortcuts →

Paper management

Link papers to chats for context-aware conversations

Collections

Organize chats into topic-based collections

Subscription tiers

Compare model access across different plans

Tips & best practices

Learn advanced strategies for using AI chat effectively