How do I use the hash-linking system in Playbooks?

Quick Answer

The hash-linking system in Jylo lets you connect different parts of your analysis by referencing outputs from other prompts. Use the # symbol to reference prompt outputs, the * symbol to reference input documents, and the @ symbol to reference source documents. This powerful feature enables sophisticated multi-stage analysis chains where later prompts can build upon earlier results.

Detailed Steps

Hash-linking is one of the most powerful features in Jylo, allowing you to create sophisticated, connected analyses across your Playbooks. Here's how to use it effectively:

1. Navigate to a prompt field while creating or editing a Playbook
2. Type the appropriate symbol based on what you want to reference:
   • Type # to reference other filter outputs
   • Type * to reference input documents
   • Type @ to reference source documents
3. After typing the symbol, a dropdown menu will appear showing available references
4. Select the reference you want to include
5. Complete your prompt with additional instructions around the reference

Understanding the Three Reference Types

# Prompt References

The # symbol lets you reference outputs from other prompts in your Playbook. This is useful for:

• Building on previous analysis
• Comparing results across different prompts
• Creating summary analyses that incorporate multiple findings

Example: "Review the following contract terms and provide a summary of key risks: ContractTerms = #ContractAnalysisFilter"

* Input Document References

The * symbol references the document currently being analysed in a Flow. This is useful for:

• Instructing the AI to look at specific sections of the document
• Creating prompts that analyse the entire document contextually
• Ensuring the AI focuses on the right content

Example: "Please extract all party names mentioned in the following document *Input Document"

The input document should be referenced in all prompts unless you are referencing responses to other prompts using the # symbol.

@ Source Document References

The @ symbol references documents you've uploaded as sources during Playbook creation. This is useful for:

• Comparing input documents against standard templates
• Checking compliance against policy documents
• Using reference materials to guide analysis
• Attaching reusable instructions, personas, or specialised knowledge (like legal perspectives)
• Creating consistent formatting guidelines that can be applied across multiple prompts

Example: "Compare the following documents and identify any discrepancies:

MainContract = *Input Document

StandardTemplate = @ContractTemplate"

Tip

When working with multiple document references in a single prompt, keep your hash-link symbols separate from your instructions by using a clear assignment format. For example:

"Please compare the contract against our issues list and highlight any potential problems. Contract = *Input Document IssuesList = @StandardIssues"

This naming approach is only necessary when using multiple references in the same prompt. For simpler prompts with a single reference, you can simply use the symbol directly.

Common Questions

What happens if a referenced prompt doesn't return a result?

For a hash reference (#) to work properly, the referenced prompt must have been successfully triggered and completed. If the prompt wasn't triggered due to conditional logic or didn't return a result, the referring prompt won't have that input to work with. For best results, use hash references primarily in follow-up prompts that logically build upon earlier analysis steps.

Where exactly should I place the reference symbols in my prompts?

When using reference symbols (* or @ or #), remember that wherever you place the symbol is where the AI will see the entire content of the referenced material. For clearer instructions, it's best to keep symbols separate from your instruction text and consider using a variable-style assignment for better readability.

If You're Still Stuck...

If you're having trouble with hash-linking:

1. Check that you're referencing prompts that appear earlier in your Playbook flow
2. Verify that the referenced prompts are likely to return results
3. Use descriptive prompt headings to make your references more intuitive
4. Test your Playbook with sample documents to see if hash-links work as expected

Still need help? Contact us via Zendesk.

Related Articles

• Setting up Conditional Logic in your Playbooks
• How to create a Playbook for effective Flows
• What are Playbooks and how do they work?
• What are the different prompt types and when should I use them?