Thinking Modes
Arketic AI assistants can think at different depths depending on how complex your question is. Choosing the right mode helps you get the best answer in the right amount of time.
Available Modes
| Mode | Speed | Depth | Best For |
|---|---|---|---|
| Fast | Instant | Surface-level | Simple facts, quick lookups, yes/no questions |
| Automatic | Varies | AI decides | General use — the AI picks the right depth for you |
| Deep | Slower | Detailed reasoning | Complex decisions, comparisons, multi-step analysis |
| Ultra | Slowest | Maximum reasoning | Very complex problems, strategic planning, thorough research |
Automatic is the recommended default. The AI analyzes your question and decides whether it needs quick or deep thinking — so you don't have to choose every time.
How to Change Thinking Mode
- Look at the chat input area at the bottom of the screen
- Find the thinking mode selector (usually shows the current mode name)
- Click it to open the mode dropdown
- Select your preferred mode
- Type and send your message
The mode you select stays active for the rest of the conversation. You can change it at any time between messages.
What Happens in Each Mode
Fast Mode
- The AI responds immediately with a direct answer
- No visible thinking process
- Best when you already know what you're looking for
Example: "What's the phone number for IT support?"
Automatic Mode
- The AI evaluates your question first
- Simple questions → fast response
- Complex questions → automatically switches to deeper thinking
- You don't see the decision process
Example: "Summarize the Q3 report" — AI decides this needs moderate depth
Deep Mode
- A thinking bubble appears before the answer
- You can see the AI's step-by-step reasoning
- The AI considers multiple angles before responding
- Takes longer but gives more thorough answers
Example: "Compare our two vendor proposals and recommend which one to choose"
Ultra Mode
- The most thorough analysis available
- Extended thinking bubble with detailed reasoning
- The AI explores multiple perspectives, considers edge cases
- Takes the longest but gives the most comprehensive response
Example: "Analyze the impact of restructuring the sales department across all regions"
Seeing the AI Think
In Deep and Ultra modes, a special thinking bubble appears while the AI works. This shows:
- The questions the AI is asking itself
- The steps it's taking to analyze your request
- The sources and documents it's referencing
- Its reasoning before reaching a conclusion
After the thinking process completes, the final answer appears below the thinking bubble.
The thinking bubble is read-only — you can't interact with it. It's there so you can understand and verify the AI's logic.
When to Use Which Mode
| Scenario | Recommended Mode |
|---|---|
| Quick factual question | Fast |
| Most everyday questions | Automatic |
| "What does our policy say about...?" | Automatic |
| "Summarize this document" | Automatic |
| "Compare X and Y and recommend" | Deep |
| "What are the risks of this decision?" | Deep |
| "Analyze the full impact of restructuring on every department" | Ultra |
| "Create a detailed action plan for the next quarter" | Ultra |
Tips
- Start with Automatic — only switch to Deep/Ultra when you need thorough analysis
- Deep and Ultra take longer — don't use them for simple questions, you'll just wait unnecessarily
- You can change modes mid-conversation — ask a quick question in Fast, then switch to Deep for a complex follow-up
- Thinking bubbles are useful for verification — read the reasoning to make sure the AI understood your question correctly