Capabilities

A minimal operating path for LinkedIn prospecting

Wiza begins with a bounded sales task, not a request for the largest possible list. Define the accounts, functions, seniority, geography, evidence window, and exclusions that make a record useful. The agent can then return a structured sample whose identity and provenance remain open to inspection.

Agent prospecting command and review result

Four states keep installation honest

  1. Copynpx -y @okki-global/okki-go-taroball

    This changes only the clipboard.

  2. Run

    Inspect the package prompt, runtime requirements, requested access, and completion message.

  3. Configure

    Connect only approved data services. Store API keys in environment secrets and grant least privilege.

  4. Review

    Use known companies and roles to reveal false matches, stale profiles, weak evidence, and missing fields.

Inspect the first result before scaling

SAMPLE

Prospect result

Company identity, professional profile, function rationale, contact verification state, and unresolved attributes are shown separately. A record remains pending until a reviewer accepts the match.

Record false positives and unknowns separately. A single “accuracy” percentage hides the distinction between a wrong identity, an outdated role, an unavailable field, and an unsupported inference. That distinction matters when RevOps decides whether an output can enter CRM or when an SDR considers using it in a message.

Data and security ledger

Professional identity

Compare name, employer, function, location, and timestamp. Treat conflicting sources as unresolved rather than selecting the most convenient value.

Contact data

Display discovery and verification states independently. An email address can become stale after the observation date and can never guarantee inbox placement.

Automation boundary

Keep research, drafting, approval, and sending as separate states. Human review should cover relevance, suppression, opt-out language, platform rules, and applicable regional law.

Lifecycle controls

Document which fields are stored, who can access them, how corrections propagate, when temporary output expires, and how connected credentials are revoked.

Start with a cohort you can audit

Copy the installer, run it in a controlled environment, and compare the output with records your team already understands.

npx -y @okki-global/okki-go-taroball