Ctrlk
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Best Practices and Workflows

This section explains how to use the platform efficiently by selecting the right model, enabling tools appropriately, structuring knowledge properly, and optimizing credit consumption. Following these workflows improves output quality, consistency, and cost efficiency.

1. Choosing the Right Model

Selecting the correct model is the first step in achieving accurate and efficient results.

1.1 Understand Your Objective

Before choosing a model, define:

  1. Is this a creative task or a logical task?

  2. Do you need deep reasoning or quick responses?

  3. Are you generating text, images, or analyzing files?

  4. Is cost efficiency more important than performance?

1.2 Match Model Type to Task

General guidance:

  1. Creative writing, marketing copy, storytelling

  2. → Choose models known for creative and expressive output.

  3. Technical analysis, reasoning, coding

  4. → Choose models optimized for structured reasoning.

  5. Image generation

  6. → Select models that support image generation.

  7. Large documents or long PDFs

  8. → Choose models with higher context length.

1.3 Compare Model Information

Before finalizing, check:

  1. Context length

  2. Supported features (text, image, reasoning)

  3. Credits per 1,000 input tokens

  4. Credits per 1,000 output tokens

Higher-performing models often consume more credits, so align model capability with task complexity.

Best Practice

Do not use the most powerful model for simple tasks.Use lightweight models for drafting, and advanced models for final output or benchmarking.

2. When to Use Auto Mode

Auto Mode automatically enables relevant tools based on your prompt.

2.1 Use Auto Mode When

  1. Your prompt may require web data.

  2. You are unsure which tool is required.

  3. You want faster workflow without manual toggling.

  4. Your task involves mixed inputs (files, URLs, search, etc.).

2.2 Avoid Manual Toggling When

Manually enabling every tool can:

  1. Increase complexity

  2. Slow down workflow

  3. Lead to unnecessary tool activation

Auto Mode simplifies decision-making by activating tools only when needed.

Best Practice

Keep Auto Mode enabled unless you need strict control over which tools are active.

3. When to Create a Knowledge Base

A Knowledge Base allows you to provide persistent reference material to the model.

3.1 Create a Knowledge Base When

  1. You frequently reference the same documents.

  2. You work with company policies or internal SOPs.

  3. You need consistent answers from structured documents.

  4. You want to avoid re-uploading files repeatedly.

3.2 Benefits of a Knowledge Base

  1. Faster responses

  2. Context consistency

  3. Reduced repetitive uploads

  4. Better accuracy for domain-specific tasks

3.3 Workflow Example

Instead of uploading a product manual every time:

  1. Create a Knowledge Base once.

  2. Attach it to your agent or use it in chat.

  3. Ask questions referencing that stored knowledge.

This improves both speed and credit efficiency.

4. When to Create a Custom Agent

Custom Agents allow you to define a persistent persona and attach knowledge resources.

4.1 Create a Custom Agent When

  1. You repeatedly perform similar role-based tasks.

  2. You need consistent tone and expertise.

  3. You want persistent system-level instructions.

  4. You require domain-specific knowledge integration.

4.2 Benefits of Custom Agents

  1. Saves time on rewriting instructions

  2. Maintains consistent output style

  3. Reduces prompt length

  4. Improves task specialization

4.3 Workflow Example

Instead of writing:Act as a senior marketing strategist with SaaS experience...Every time, create a custom agent with:

  1. Defined role

  2. Expertise description

  3. Attached knowledge base

Then simply start working within that agent.

5. Optimizing Credit Usage

Efficient credit usage ensures long-term sustainability and cost control.

5.1 Choose the Right Model Tier

  1. Use lightweight models for drafts.

  2. Use premium models for final outputs.

  3. Avoid high-cost models for simple tasks.

5.2 Write Clear Prompts

Unclear prompts cause:

  1. Multiple iterations

  2. Rewrites

  3. Extra token consumption

Clear prompts reduce unnecessary back-and-forth.

5.3 Manage Context Length

Long chat threads increase token usage because previous messages remain in context.Best practice:

  1. Start a new chat for new objectives.

  2. Reset context when switching topics.

5.4 Avoid Redundant Attachments

  1. Use Knowledge Base instead of repeatedly uploading files.

  2. Avoid attaching large documents unless required.

5.5 Control Output Length

If you do not need a long response:

  1. Specify word limit.

  2. Ask for summary instead of full expansion.

Longer outputs consume more output tokens and credits.

6. Recommended Workflow Summary

  1. Define your objective clearly.

  2. Select the appropriate model based on task complexity.

  3. Enable Auto Mode unless strict control is required.

  4. Create a Knowledge Base for recurring documents.

  5. Build Custom Agents for repeated role-based tasks.

  6. Use clear prompts and reset context when needed.

  7. Monitor token usage and output length to optimize credits.

Following these structured workflows ensures:

  • Higher quality outputs

  • Faster task execution

  • Better organization

  • Efficient credit management

Last updated