Controlled prospecting process from brief to review

Capabilities

Seamless AI turns a research brief into a review queue

Each stage keeps its inputs, evidence, exceptions, and acceptance decision visible so operators can correct the process instead of trusting a filled field.

Four controlled research services

01

Company discovery

Translate industry, stage, geography, and exclusions into a repeatable query. Return operating identity, domain, location, and dated context with unknowns intact.

02

Entity resolution

Compare legal name, operating name, domain, headquarters, and event clues. Route collisions and ambiguous subsidiaries to review rather than merging silently.

03

Decision-maker research

Map functions and likely responsibilities without treating a title as proof of authority or current employment. Preserve source date and verification state.

04

Outreach preparation

Assemble facts, inferences, open questions, suppression state, and regional requirements into a draft that waits for a named human approver.

A numbered method with explicit exits

  1. Define

    Write acceptance criteria, proxies, exclusions, and a freshness window before seeing results.

  2. Resolve

    Confirm company identity before enriching role or event fields.

  3. Evidence

    Attach source context and collection date; leave unsupported fields unknown.

  4. Review

    Approve fit, relevance, suppression, opt-out language, and regional policy before outreach.

What the workflow does not guarantee

Record availability, match quality, title currency, inbox placement, consent, replies, and commercial fit vary by source, market, configuration, and reviewer judgment. Test on a known cohort and document failures before expanding scope.

A useful acceptance log records false merges, unresolved domains, stale employment evidence, conflicting titles, and unsupported intent in separate categories. Reviewers can then decide whether to narrow the brief, change a source, adjust a freshness window, or leave a field unresolved. This makes improvement measurable without inventing a universal accuracy percentage.

Run the method on accounts you already know

Install the skill, define one acceptance sheet, and compare returned evidence with your verified reference set.

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