Senso
Sign inSign up

Context Layer · 5 of 5

Topics

How to tag things, so a gap can find the source that should fix it.

The five sections

Brand KitVoice, tone and the rules generation inheritsKnowledge BaseThe documents every answer is built fromContent TypesTemplates for the things you publishProduct CatalogWhat you sell, in a form an agent can quoteTopicsThe subjects you want to be found forYou are here

Endpoints

Open in API reference →
get/org/tagsEvery topic, with countspost/org/tagsCreate a topicpatch/org/tags/{id}Rename a topicput/org/kb/nodes/{id}/tagsSet a document’s topicsput/org/prompts/{promptId}/tagsSet a prompt’s topicsput/org/content/{id}/tagsSet a draft’s topics

What they are for

A topic is a label attached to prompts, knowledge base nodes, content items and search queries. It is how a gap found while evaluating lines up with the source that should fix it and the content that should carry it.

Funnel stage

Awareness, Consideration, Evaluation, Decision.

Product area

Knowledge base, Publishing, Analytics.

Customer intent

Pricing, Setup, Troubleshooting.

Jurisdiction

US, Canada, EU.

Funnel stage matters most. It is what lets you say “we lose the decision-stage questions” rather than “we lose some questions” — and decision-stage gaps are the ones worth closing first.

Naming

The docs call these Topics because that is what they do. The API calls them tags, and the endpoints are /org/tags.

You mostly do not create these by hand

The library grows on its own.

Senso auto-tags on creation, for every taggable resource:

  • Prompts, on POST /org/prompts
  • Raw content, on POST /org/kb/raw
  • Uploaded files, once ingestion finishes
  • Search queries, as they arrive on /org/search*

Query tagging classifies on intent — “how do I update my beneficiary” becomes beneficiary updates — which populates query analytics without anyone doing anything.

Attaching by name

bash
# Setting topics replaces what was there. Unknown names are created.
curl -X PUT https://apiv2.senso.ai/api/v1/org/kb/nodes/{id}/tags \
  -H "X-API-Key: $SENSO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tag_names": ["Pricing", "Decision"]}'
  • Names are unique per organization, case-insensitive.
  • Setting topics replaces what was on the item rather than adding to it.
  • Topics apply to documents, not folders. Folders organize the tree; topics cut across it.

Counts tell you where the gaps are

bash
# Every topic, with where it is used
curl "https://apiv2.senso.ai/api/v1/org/tags?counts=true" \
  -H "X-API-Key: $SENSO_API_KEY"

Counts come back across prompts, content, knowledge base content and generated content. A topic with prompts but no content is a gap you have not filled yet — which is exactly what module 2 is for.

Where they get used

  • Module 1 — applied as you ingest, mostly automatically
  • Module 2 — gaps are prioritized by consequence and funnel stage
  • Module 4 — metrics roll up by topic, so you can see which part of the funnel moved
Build the Context LayerAnalyticsAPI reference
Back to module 1