Search and filtering

Ultrathink offers several ways to find entries in your knowledge base: a search bar for text lookups, AI search for natural-language questions, and sidebar filters for browsing by type, entity, and category.

The search bar is located above the main grid on the All entries page.

How it works

  1. Type a keyword or phrase into the search bar.
  2. Results filter as you type, matching against entry titles and content.
  3. Press Enter (with at least three characters) to run an AI-powered search for more relevant results.

Features

  • Instant filtering: results narrow as you type.
  • AI search on Enter: pressing Enter triggers semantic search for richer results.

For more complex queries, use the dedicated AI search page.

Click Search in the sidebar to open the AI search interface.

How it works

  1. Type a natural-language question in the search box (for example, "What did I save about React performance?").
  2. Ultrathink's AI interprets your intent.
  3. Semantic search finds relevant entries based on meaning, not just keywords.
  4. An AI-generated answer is displayed at the top, synthesised from your own content.
  5. Matching entries appear below with highlighted context.

Example queries

QueryWhat it finds
"What did I save about React hooks?"React-related entries with hook content
"Meeting notes from last week"Recent entries classified as meetings
"John's recommendations"Entries mentioning John that contain advice
"Project deadlines"Tasks and projects with due dates
"That article about serverless"Serverless-related links and snippets

Filtering AI search by entity

Narrow your AI search results to a specific entity type:

  • All: search everything.
  • Projects: only project entries.
  • Tasks: only task entries.
  • Knowledge: only knowledge entries.

Select the entity filter before or after running your query.

AI-generated answers

The AI composes a direct answer from your entries, complete with references to the source material. For example:

Q: "What were the key decisions from the product meeting?"

A: "In your 15 March meeting notes, three key decisions were made: (1) postpone the mobile app launch to Q3, (2) prioritise API documentation, (3) hire two more engineers by end of month."

Click any referenced entry to view it in full.

The sidebar provides quick filters with live counts so you can narrow the grid in a single click.

By type

Each content type is listed with the number of matching entries:

  • Links
  • Screenshots
  • ChatGPT conversations
  • PDFs
  • Claude conversations
  • Videos
  • And more

Click a type to filter the grid. Click again to clear.

By entity

  • Project
  • Task
  • Knowledge
  • Unclassified

By category

  • Work
  • Personal

Sidebar filters apply immediately and combine with each other.

Sorting

Change the order of results using the sort control in the toolbar.

Sort optionDescription
Date created (newest)Most recently captured first
Date created (oldest)Earliest first
Date updatedMost recently modified first
Title (A-Z)Alphabetical order
Title (Z-A)Reverse alphabetical order
TypeGrouped by content type

Sorting works alongside active filters.

Search tips

Be specific

More specific queries return better results, especially with AI search.

Instead ofTry
"react""React performance optimisation techniques"
"meeting""Product team meeting notes March"
"john""John's feedback on the API design"

Use natural questions

AI search responds best to full questions:

  • "What did I learn about caching strategies?"
  • "How does the authentication flow work?"
  • "Where are the notes about the Q2 roadmap?"
  • "Who suggested the new onboarding design?"

Search AI conversations

Conversations from ChatGPT, Claude, and other AI tools are fully searchable:

  • "What did Claude say about database indexing?"
  • "ChatGPT conversation about regex patterns"
  • "AI suggestions for the landing page copy"

Troubleshooting

No results found

  • Try shorter or simpler queries.
  • Check for typos.
  • Remove active filters and search again.
  • Use different terms for the same concept.

Irrelevant results

  • Add more context to your query.
  • Use entity filters to narrow the scope.
  • Check that entries are properly classified (correct type and topics).
  • AI search typically takes two to five seconds for complex queries.
  • Text search filtering is near-instant for simple lookups.
  • Very large knowledge bases may take slightly longer.

Next steps