Lead intelligence
Ask for bounded account and contact research. Keep source, freshness, confidence, suppression, and review state attached to the result.
See the research handoff ↗Product / the operating layer
One workspace for the outputs your agent needs: evidence-backed leads, sequence previews, mailbox readiness, and reply-driven next actions.
Walk through the system ↗Agents prepare the evidence, state, and next action in one place.
People handle policy, exceptions, and the final decision—not every click.
Returned capability and source freshness stay visible before action.
01 / Four connected surfaces
Humans set intent and approve the edge. The agent carries the repeatable middle through one inspectable contract.
Ask for bounded account and contact research. Keep source, freshness, confidence, suppression, and review state attached to the result.
See the research handoff ↗Turn approved human memory into sequence drafts, validate variables, and preview what would happen before an apply action.
See campaign preparation ↗Make sender state, limits, suppression, and approval requirements visible before a message can leave the system.
See safety boundaries ↗Classify replies, stop sequences where needed, and hand a clear, reviewable next action back to the team.
See the reply loop ↗05 / Evidence
Evidence receipts give humans and verifier agents a way to inspect what changed, what was allowed, and what still needs attention.
06 / Source-bound preview
This read-only sample uses the same names and state vocabulary exposed by the current product contract. It is sanitized and cannot send, connect a provider, or change a workspace.
Routes, scopes, idempotency, approval references, and receipt states are documented in the agent contract. Values in this panel are not a customer account or performance claim.
Open the evidence boundarySee it in context