Use cases by operating context
Coverage, communication rules, source access, and review expectations differ by market and team. Start with the operating context instead of an abstract record count.
Test a market thesis using explicit industry boundaries, geography, company-stage proxies, event recency, and exclusions before judging an account list. Do not widen the criteria after seeing a convenient result.
Acceptance question: Would the account still fit if a single event signal disappeared?
Resolve company identity before enriching routing fields. Preserve timestamps, transformation rules, fallback order, and an explicit unknown state when providers disagree.
Acceptance question: Can an operator explain and reverse every field change?
Prepare short account briefs with dated evidence, role rationale, and open questions. Keep research, drafting, approval, and sending as separate states.
Acceptance question: Can a representative distinguish fact, inference, and unsupported personalization?
Evaluate runtime permissions, secret storage, rate limits, output schema, retention, deletion, and safe failure behavior with a known cohort.
Acceptance question: Can the workflow stop safely when identity evidence is weak?
Decision controls
Browser automation can expose useful context while creating fragile permission boundaries. API access can be repeatable while adding rate-limit and schema dependencies. Waterfall enrichment may raise field coverage while increasing conflict and deletion work. Measure the trade-off in the workflow you actually run.
Install the workflow, run a small test, and assess entity identity, dated evidence, unresolved fields, human review, and applicable communication controls. Record the reviewer, test date, source exceptions, and every rejected result.
npx -y @okki-global/okki-go-taroball