Timeline Reply Summary

The Timeline Reply Summary function uses AI to automatically generate concise summaries of timeline post replies. Similar to Reddit’s TL;DR feature, it helps users quickly understand the key points of a discussion without reading every reply.

Overview

When a timeline post accumulates enough replies, an AI-generated summary becomes available. The summary appears as a collapsible section when users expand the replies, providing a quick overview of the discussion’s main points, common themes, and any notable disagreements.

Key Features

• AI-generated summaries of timeline reply discussions
• Support for OpenAI and Anthropic AI providers
• Configurable reply count threshold to trigger summaries
• Customizable AI prompt template
• Summaries cached in database until replies change
• Collapsible UI that appears only when replies are expanded
• Works on reviews, post walls, post timelines, and user timelines
• Lazy loading – summary fetched only when user clicks to view

Setup Instructions

Enable the Function

1. Navigate to Voxel Toolkit → Functions in your WordPress admin
2. Find Timeline Reply Summary in the list
3. Toggle the switch to enable the function
4. Click Save Changes

Configure AI Provider

1. Click the AI Settings Tab under Settings
2. Select your AI Provider (OpenAI or Anthropic)
3. Enter your API Key for the selected provider
4. Click Save Changes

Important: You must provide a valid API key for the feature to work. API usage will be billed by your AI provider based on their pricing.

Settings

Setting	Description	Default
AI Provider	Choose between OpenAI or Anthropic for AI generation	OpenAI
API Key	Your API key for the selected AI provider	–
Reply Threshold	Minimum number of replies before a summary is generated	3
Max Summary Length	Maximum tokens for the AI-generated summary	300
Label Text	Text displayed on the summary toggle button	TL;DR
Enabled Feeds	Which timeline feeds should show summaries	All feeds
Prompt Template	Custom AI prompt for generating summaries	Default prompt

How It Works

Summary Generation

1. A timeline post receives replies from users
2. When the reply count reaches the threshold, the post becomes eligible for a summary
3. The summary is generated when a user first expands it (lazy loading)
4. All approved replies are sent to the AI with your prompt template
5. The AI generates a concise summary of the discussion
6. The summary is cached in the database for future views

Summary Regeneration

Summaries are automatically regenerated when:

• New replies are added to the post
• Replies are deleted or moderated
• The reply count changes from when the summary was generated

User Experience

1. User views a timeline post with multiple replies
2. User clicks the reply count to expand replies
3. If eligible, a “TL;DR” button appears above the replies with an “AI” badge
4. User clicks the button to expand and view the summary
5. Summary loads (or generates if first view)
6. When replies are collapsed, the summary hides automatically

Supported Feeds

Reply summaries work on all Voxel timeline feeds:

Feed Type	Description
Post Reviews	Reviews left on listings
Post Wall	Wall posts on listing pages
Post Timeline	Timeline posts on listings
User Timeline	User profile timelines and newsfeed

You can enable or disable specific feeds in the function settings.

AI Providers

OpenAI

• Model: GPT-4o-mini
• Get API Key: platform.openai.com/api-keys
• Pricing: Pay-per-use based on tokens

Anthropic

• Model: Claude 3 Haiku
• Get API Key: console.anthropic.com
• Pricing: Pay-per-use based on tokens

Both providers offer fast, cost-effective models suitable for summarization tasks.

Custom Prompt Template

You can customize the AI prompt used to generate summaries. The default prompt is:

Summarize the following discussion replies concisely in 2-3 sentences.
Focus on:
- Main points and key takeaways
- Any consensus or common themes
- Notable disagreements or different perspectives

Keep the tone neutral and informative. Do not include usernames.

Replies:
{{replies}}

Summary:

Use the placeholder where you want the reply content to be inserted. The replies are formatted as a bulleted list of quoted text.

Prompt Tips

• Be specific about the summary length you want
• Specify the tone (neutral, friendly, professional)
• Tell the AI what to focus on or ignore
• Request specific formatting if needed

Database Storage

Summaries are stored in a dedicated database table (wp_voxel_toolkit_reply_summaries) with the following information:

• Status ID (linked to the timeline post)
• Generated summary text
• Reply count at generation time
• Generation timestamp
• AI provider used

This caching ensures fast load times and minimizes API costs by only regenerating when replies change.

Use Cases

Review Aggregation

Quickly understand the overall sentiment and key points from multiple reviews on a listing.

Discussion Summary

Catch up on long discussion threads without reading every reply.

Community Insights

See what your community is saying about topics at a glance.

Time Saving

Users can quickly decide if they want to read the full discussion based on the summary.

Troubleshooting

Summary Not Appearing

• Check that the function is enabled
• Verify your API key is entered correctly
• Ensure the post has enough replies (check threshold setting)
• Make sure the feed type is enabled in settings
• Click to expand replies first – summary only shows when replies are visible

Summary Shows Error

• Check your API key is valid and has credits
• Verify your API account is in good standing
• Check server error logs for API response details
• Try switching AI providers if one isn’t working

Summary Not Updating

• Summaries regenerate when reply count changes
• New replies must be approved/moderated first
• Clear browser cache and refresh the page

Summary Not Hiding

• The summary should hide when replies are collapsed
• Clear browser cache and ensure you have the latest version
• Check for JavaScript errors in browser console

API Cost Considerations

To minimize API costs:

• Set an appropriate reply threshold (higher = fewer summaries)
• Summaries are cached and only regenerate when replies change
• Lazy loading means API is only called when users view summaries
• Both GPT-4o-mini and Claude 3 Haiku are cost-effective models
• Long replies are truncated before sending to the API

Requirements

• OpenAI or Anthropic API key
• Timeline features in use on your site