Signals
The organization's interpretation layer, extracted from conversations and attached to records.
A transcript preserves the raw record. A summary makes it readable. Signals capture the interpretation: the judgments, intent, constraints, risks and commitments that explain why an outcome happened rather than just what it was.
What they represent
Signals are the things worth keeping from a conversation:
- Decisions and outcomes: what was agreed, approved, rejected or deferred.
- Rationale: why, and what trade-offs were accepted.
- Intent and priorities: what the customer or the team is optimising for.
- Constraints and requirements: what the answer has to live within.
- Risks and blockers: objections, dependencies, unresolved questions.
- Commitments: who owns what, and what finished looks like.
What makes something a signal is that your organization decided it matters enough to record, track and act on. The catalog is yours, not a generic one.
Why they matter
They keep the reasoning. Most systems record the outcome and lose the judgment behind it. A signal keeps both, with the quote it came from.
They make people consistent. Left to free-form notes, every person captures something different. A defined signal catalog means the same thing is captured the same way across a team.
They accumulate. Signals build up on a record, so someone new can understand where an account stands without replaying six months of calls.
What a signal carries
A category and a name from your catalog, so signals are comparable and reportable. The record it belongs to. Provenance back to the conversation, and the verbatim quote it was drawn from. A confidence score. Enough structure to hold the interpretation explicitly. And a lifecycle, so a signal can be worked without its history being erased.
The quote is the part to read first. It is the evidence, and it is what lets a person judge whether the interpretation above it is fair.
Aggregating
Because signals are structured and attached to records, they group cleanly: by record, to see the state of one account; by category, to see recurring blockers or decision drivers; by team, region or segment, to see patterns without losing the detail; and across time, to see what changed and what is still unresolved.
Where this lives in the product
Signal definitions and categories are configured in the console under Signals, and attached to an agent before it will detect anything. Detections appear inside each session, on the organization-wide signals screen, in Analytics, and on the phone as a session's Signals tab. The rules governing how signals are extracted and reviewed live in the AI constitution.