Evidence-led company researchHuman review before outreach

Auditable B2B company discovery

Build a prospect list your revenue team can actually review

Describe the market, firmographic criteria, exclusions, and evidence window. UpLead turns that brief into a structured company shortlist while keeping domains, source dates, uncertainty, and human approval visible.

See the evidence checks
Agent terminal with an auditable company discovery result

Install in your AI agent

npx -y @okki-global/okki-go-taroball
  • Brief accepted
  • Company identities resolved
  • Unknown fields preserved

Evidence before volume

Four controls for dependable prospect research

01

Define the addressable market

Translate an ICP into observable firmographic and technographic filters, an explicit geography, a recency window, and disqualifying conditions.

02

Resolve company identity

Compare operating name, legal identity, primary domain, location, and dated events before records are treated as the same organization.

03

Verify decision-maker context

Separate role relevance from contact availability. Keep email verification state, title ambiguity, and missing evidence visible to the reviewer.

04

Approve the next action

Require a human-in-the-loop gate before enrichment changes a CRM or a draft enters an outreach sequence.

4 statesCopy, run, configure, review
3 checksIdentity, recency, provenance
1 gateHuman approval before outreach
0 promisesNo guaranteed match, inbox, or reply

Data behavior ledger

Trust comes from inspectable states

Confirm these controls in your own environment. A field is useful only when its origin, observation date, transformation, and review status remain understandable.

Config-dependent

Data sources

Document connected providers, permitted use, fallback order, and field-level provenance. Waterfall enrichment may improve coverage while increasing conflicts and deletion work.

Verified practice

API keys

Keep credentials in a runtime secret store with minimum scope. Never place keys in prompts, source control, shared logs, or exported prospect records.

Inspect locally

Runtime behavior

Review package permissions, network destinations, API rate limits, retry behavior, temporary files, and failure output before production use.

Config-dependent

Retention and deletion

Define access, correction, suppression, retention, and deletion for intermediate results as well as final CRM records.

Installation runbook

Four states, four different success signals

  1. 01 · Copynpx -y @okki-global/okki-go-taroball

    Success means the exact command is on your clipboard. Nothing has been installed yet.

  2. 02 · Run

    Execute it in a supported terminal, read the package prompt, and retain the package version and sanitized completion output.

  3. 03 · Configure

    Add only required credentials through environment secrets. Confirm permission scope and connected data terms before the first request.

  4. 04 · First result

    Use a known cohort containing good fits, exclusions, ambiguous domains, and stale events. Review each claim before any downstream action.

Practical answers

Questions to settle before prospecting

Installation health and research quality are separate. Validate both before connecting a live outreach workflow.

Does copying the command install the skill?

No. Copying changes the clipboard only. Installation begins when an operator runs the command and reviews the package prompt in a terminal.

Where does company and contact data come from?

That depends on the sources configured in the runtime. Record provider, purpose, field provenance, observation date, and usage restrictions for every connection.

Does a verified email guarantee delivery?

No. Verification is a point-in-time technical signal. Sender reputation, SPF, DKIM, DMARC, mailbox behavior, content, and recipient policy still affect delivery.

Can the agent send outreach automatically?

Keep sending disconnected during evaluation. Use human review for targeting, claims, relevance, suppression, opt-out handling, and applicable regional requirements.

How should UpLead be compared with a seat-based platform?

Use the same known cohort and measure accepted identities, unresolved fields, evidence freshness, review time, correction effort, governance controls, and total operating cost.