Good prospecting leaves room to say “not enough evidence”
Kaspr’s working culture treats uncertainty, correction, and human judgment as product behavior rather than friction to hide.
Mission
Shorten research without shortening review
Help sales operators move from a written ICP to a structured prospect record inside the AI agent they already use. Preserve identity clues, role rationale, source context, dates, exclusions, and unresolved fields so the result can be challenged.
Operating view
Evidence before activation
Company discovery, profile interpretation, email verification, drafting, approval, and sending are distinct decisions. The workflow should make those boundaries visible and allow a reviewer to stop, correct, or delete a record.
Review habits built into the work
01
State the unknown
An empty or conflicting field stays unresolved instead of becoming a confident-looking inference.
02
Keep provenance
Important claims retain their source context, observation date, and transformation history.
03
Minimize access
Credentials use the smallest practical scope and never belong in prompts or shared logs.
04
Separate sending
A person reviews relevance, suppression, opt-out, regional rules, and message claims.
Four working surfaces
Brief and acceptance criteriaRole and recency reviewLeast-privilege setupApproval before outreach
Bring a known cohort, not a vanity benchmark
Test positive matches, exclusions, stale roles, ambiguous companies, and missing evidence. Record why each result passes or fails before expanding the task.
A useful review note names the accepted identity, rejected alternative, source context, observation date, role rationale, suppression status, and downstream consequence. It also distinguishes a missing field from a failed match. That discipline gives another operator enough information to reproduce the decision, correct it later, or delete data that no longer serves the stated purpose.
When the workflow supports outreach preparation, operators should verify sender authentication, applicable regional requirements, platform terms, opt-out handling, and a named human approver. No cultural principle, interface, or automation step can guarantee consent, inbox placement, replies, or commercial fit.