Auto-Tagging

Manually tagging every note is tedious, especially as your vault grows. Ariv’s auto-tagging system handles this for you using a two-tier approach: a free, local heuristic layer that always runs, and an optional AI refinement layer that adds deeper understanding.

How It Works

When you create or edit a note, Ariv analyzes its content and suggests tags. These suggestions appear in the footer of the editor, below your note content. From there, you can accept, dismiss, or edit any suggested tag before it’s applied.

Tags are never forced on you — Ariv suggests, you decide.

Tier 1: Heuristic Tagging (Free, Local, Always On)

The first tier runs entirely on your machine with no API calls and no cost. It uses three complementary strategies to generate tag suggestions:

Vault Propagation

Ariv learns from the tags already in your vault. If many of your notes about “machine learning” are tagged #ml, Ariv picks up on that pattern and suggests #ml for new notes with similar content.

The more consistently you tag, the smarter propagation becomes. It’s a virtuous cycle — each tagged note teaches Ariv a little more about your personal vocabulary.

TF-IDF Analysis

TF-IDF (Term Frequency — Inverse Document Frequency) identifies the most distinctive terms in a note relative to the rest of your vault. Words that appear frequently in one note but rarely across your vault get surfaced as tag candidates.

This catches the specific topics a note is about, rather than common words that appear everywhere.

Heading Extraction

Section headings in your notes are strong signals for what a note covers. Ariv uses ## and ### headings as additional tag candidates, since headings tend to concisely describe the content that follows.

Tip

Tier 1 heuristic tagging gets better over time. The more notes you tag — whether manually or by accepting AI suggestions — the richer your vault’s tag vocabulary becomes, and the more accurate vault propagation gets. Early effort pays off.

Tier 2: AI Refinement (Requires API Key)

When you have an AI provider configured, Ariv can use a language model to refine and improve tag suggestions beyond what heuristics alone can achieve.

What AI Adds

• Contextual understanding: The AI grasps the meaning of your note, not just its keywords. A note about “increasing team velocity” might get tagged #engineering-management even if those words never appear.
• Tag confirmation: AI reviews heuristic suggestions and confirms or rejects them, reducing noise.
• Additional suggestions: The AI may suggest tags that heuristics missed entirely, based on the semantic content of the note.

How It Works

1. Ariv takes the content of your note (truncated to roughly 5,000—8,000 characters for efficiency).
2. The content is sent to your configured AI provider along with the heuristic suggestions.
3. The AI returns a refined set of tags — confirmed, rejected, or newly suggested.
4. Refined tags appear in the editor footer alongside heuristic suggestions.

Limits and Requirements

• Minimum content: Notes need at least 20 characters of content before AI tagging runs. Very short notes don’t have enough signal.
• Maximum tags: Ariv caps at 10 tags per note to keep things focused and useful.
• Content truncation: Only the first ~5,000—8,000 characters are sent to the AI. For very long notes, the beginning of the content is used.
• Usage limits: Free tier includes 50 AI tag refinements per month. Ariv Pro removes this limit.

Note

AI tagging sends note content to your configured cloud provider (Gemini, OpenAI, or Anthropic). If you’re using Ollama, everything stays local. See AI Providers for privacy details on each option.

Accepting and Managing Tags

When Ariv suggests tags, they appear in the tag area in the editor footer:

• Accept a tag to apply it to the note.
• Dismiss a tag to remove it from the suggestions.
• Edit a tag before accepting to adjust the wording.

Accepted tags become part of the note’s metadata and are used across Ariv for:

• Search: Filter notes by tag in the sidebar or search dialog.
• Ask Brain: Tags help Ariv find relevant context when answering your questions.
• Vault propagation: Accepted tags feed back into Tier 1, improving future suggestions.

Tips for Better Auto-Tagging

1. Tag a few notes manually first. Even 10—20 manually tagged notes give vault propagation a solid foundation to learn from.
2. Use descriptive headings. Headings like ## Q3 Budget Review give the heuristic engine much better signal than ## Notes.
3. Accept good suggestions quickly. Each accepted tag makes future suggestions more accurate.
4. Don’t overthink it. Tags don’t need to be perfect. Even rough tags are useful for search and Ask Brain context.

Caution

If you rename or reorganize tags across your vault, it may take a moment for Ariv’s heuristic engine to catch up. The index updates as notes are modified, so changes propagate as you work.

---

Related: AI Setup — configure an AI provider for Tier 2 tagging | AI Providers — compare provider options | Semantic Search — another way AI enhances note discovery