Before Sedric checks your content against a single rule, it works out what the content actually is: the channel it belongs to, the regulation that governs it, what happened inside it. That understanding is called context, and the Context tab is where you can see it.
Why context matters
Not every rule applies to every piece of content. A rule about a disclaimer required in one territory shouldn't be evaluated against content published somewhere else. A rule about a payment plan is only relevant if payment actually came up.
Context is how Sedric decides. It reads the content, records what it found as attributes, and uses those attributes to determine which rules apply here at all. Only the applicable rules are then evaluated.
This is why the Context tab is the first place to look when a result surprises you. If a rule was evaluated when you didn't expect it — or skipped when you did — the answer is almost always in the context, not the rule.
Two kinds of attributes
The panel separates what Sedric worked out from what your own systems told it. The distinction matters, because it tells you who to go to when something is wrong.
Type | Where it comes from | When it's wrong |
AI attributes | Calculated by Sedric from the content itself. | Correct it here. Your correction feeds back into the AI. |
Integration attributes (metadata) | Provided by your integration — not AI-generated. These come from the system that sent the content to Sedric. | The value came from your side, so the source system is usually where it needs fixing. |
What you see depends on the content. A marketing asset might carry attributes such as its channel and the relevant regulation. An interaction might carry many more — what was discussed, whether an objection was raised, how the call ended.
Seeing where an attribute came from
Attributes aren't asserted without evidence. Where an attribute was detected at a specific point in the content, a numbered marker connects the two — the same number appears on the content itself, next to the wording or moment that produced it.
So when you want to know why Sedric decided a call involved a dispute, or that an asset belongs to a particular channel, you can go straight to the place in the content that says so.
Expand an attribute to see more about it.
Correcting an attribute
If an attribute is wrong, change it. Select the attribute and choose the correct value.
This is worth doing even when the result looks fine, for two reasons:
It fixes the reasoning, not just the record. Context drives which rules apply, so correcting it puts the analysis on the right footing.
It teaches the AI. Corrections on context feed the same feedback loop as corrections on rules and findings, and improve how future content is analyzed.
Add your reasoning whenever you're asked for it. A correction with an explanation is far more useful to the model than a silent change.
Working with a long list of attributes
Some content carries a lot of context — interactions in particular. Two tools at the top of the panel keep it manageable:
Search — jump to the attribute you're looking for by name.
Sort — reorder the list so what you care about comes first.
For any questions, feel free to contact your customer success representative or reach out via the in-app support.