Taxonomy: classify your insights your way
Insights pile up fast. Without a consistent way to categorize them, every team ends up with an informal system that lives in someone's head. This release adds taxonomy classification to Discovery. You define the groups and values that matter to your team. The AI proposes labels for every insight. High-confidence results can apply automatically; anything uncertain goes to a review queue where you accept or override. The result is a labeled dataset that matches your vocabulary.
What changed
- Taxonomy manager: define custom groups and values for classifying insights
- AI classification with confidence-based auto-apply and a human review queue
- Smart suggestions in sidebars: related opportunities and ideas for each insight
- Sidebar and navigation improvements throughout Discovery
Taxonomy classification
Open the Taxonomy Manager, define your groups (e.g. problem area, customer segment, severity), and add options for each. Run a classification batch. Zentrik proposes a label for each insight with a confidence score. Above the auto-apply threshold, labels apply immediately. Below it, they wait in a review queue where you can accept, reject, or override.
- Your team's vocabulary, not a generic default category set
- Review focuses time on the ambiguous cases, not every row
- Taxonomy labels appear as filters in discovery tables and sidebars
Smart suggestions in sidebars
When you open an insight, the sidebar now shows a Smart Suggestions tab with opportunities and ideas that are semantically related to that insight. These are not links you made manually — they are the closest matches in your workspace based on content similarity.
- Faster way to spot which opportunities an insight might already belong to
- Surfaced without search so you see connections you wouldn't have thought to look for
AI suggestion tooltip
Any field that was classified or suggested by AI shows a small indicator with the model's confidence and reasoning. Hover to see why a particular label was proposed. After you accept or override, the indicator clears.
- You know which values came from a model and which you set yourself
- The reasoning is readable before you commit to the label
More in this release
Additional updates
- Taxonomy manager with groups, options, and batch classification
- Review queue for uncertain classifications with accept/reject/override
- Smart suggestions tab on insight sidebars
- AI suggestion tooltip with confidence and reasoning
- Taxonomy distribution charts in analysis overview
- Sidebar improvements across Discovery: layout, wrapping, overflow handling
- Discovery search covers all entity types