ICP acceptance sheet
Record required company attributes, acceptable proxies, exclusions, and a freshness window.
Use with the runbookAI sales prospecting · evidence first
ZoomInfo turns a company brief into a reviewable prospect record: entity clues, dated observations, open questions, and a clear human approval gate. It does not treat a funding event, title, or web visit as automatic intent.

npx -y @okki-global/okki-go-taroballCopy → Run → Configure → First result
Method comparison
| Control | Agent-native workflow | Manual browsing |
|---|---|---|
| Brief | Reusable ICP criteria and exclusions | Analyst interprets notes per search |
| Entity resolution | Domain, name, location, and event clues compared | Usually resolved tab by tab |
| Evidence | Requested beside each returned claim | Depends on note-taking discipline |
| Unknown fields | Remain explicitly unresolved | May be omitted or inferred |
| Review | Required before any downstream outreach | Built in, but rarely standardized |
Auditable preview
Research worksheets
Record required company attributes, acceptable proxies, exclusions, and a freshness window.
Use with the runbookCompare legal name, operating name, domain, headquarters, and dated events before merging.
Read the methodologyCapture relevance, contact verification, suppression status, opt-out language, and reviewer decision.
See role-specific useInstallation runbook
npx -y @okki-global/okki-go-taroballClipboard success is not installation success.
Execute in a supported terminal and inspect the package prompt and completion output.
Store connected API credentials in environment secrets with the minimum required scope.
Test known companies; check identity, dates, source context, exclusions, and unknown states.
Questions before a first run
No. Copying changes only your clipboard. Run the command in a supported terminal, inspect the package prompt, then confirm the completion output.
Use a runtime secret store with minimum scope. Never paste API keys into prompts, exports, shared logs, or a test record.
No. Check identity, freshness, professional relevance, suppression, opt-out handling, and applicable regional requirements first.
Use known companies, one exclusion, and one ambiguous identity. Measure evidence quality and correction work rather than assuming a universal match rate.