Runtime: user-controlled · Data status: source-dependentCopy · Run · Configure · Review

CRM enrichment and buying-signal research

Warmly turns account signals into evidence your team can review

Warmly helps an AI agent assemble company identity, CRM context, visitor activity, and dated buying signals without presenting inference as fact. Every shortlist still passes through a human review gate.

See the four-state runbook
Warmly evidence-led account research workspace
npx -y @okki-global/okki-go-taroball

Next state: run and inspect the package prompt.

Six reviewable capabilities

Keep the evidence attached to the decision

Resolve accounts

Compare domain, company name, location, and CRM identifiers before merging a visitor or event into an account record.

Enrich CRM context

Request firmographic and technographic fields with source, observation date, and an explicit unknown state.

Review visitor signals

Treat website activity as an account-level clue whose identity confidence and time window require validation.

Frame buying signals

Separate observed events such as hiring or leadership change from the unverified conclusion that a purchase is imminent.

Map relevant roles

Use function and responsibility evidence to suggest reviewers; do not assume a title proves ownership or consent.

Prepare approved context

Send only reviewed facts into a draft, with suppression, opt-out, regional rules, SPF, DKIM, and DMARC considered before outreach.

4 statesCopy · Run · Configure · First result
3 evidence checksIdentity · Recency · Provenance
1 approval gateHuman review before outreach
0 guaranteesCoverage, delivery, and replies vary

Runtime matrix

Know what enters, what runs, and where results go

Input sources

CRM recordsPermission: selected account fieldsConfig-dependent
Website activityPermission: chosen event streamConfig-dependent

Research agent

Entity resolverCompares names, domains, and locationsInspect locally
Evidence ledgerRetains dates, source context, and unknownsReview required

Output destinations

Review queuePermission: approved workspace onlyRecommended
CRM updatePermission: field-level write scopeConfig-dependent

Connection names describe operating roles, not verified partnerships. Confirm provider permissions, retention, and deletion behavior before production use.

Installation runbook

Copy, run, configure, first result

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

    Success: the exact command is in your clipboard. Nothing is installed yet.

  2. 02 Run

    Execute in a supported terminal. Inspect the package prompt, requested permissions, network calls, and completion output.

  3. 03 Configure

    Keep API keys in a runtime secret store with minimum scope. Record connected sources, destinations, retention, and deletion.

  4. 04 First result

    Use known accounts. Check identity, dates, provenance, unknown fields, exclusions, and reviewer decision before any downstream action.

Categorized FAQ

Questions to settle before the first production run

Installation

Does copying the command install anything?

No. Copy only changes the clipboard; run and configuration are separate states.

Where should credentials live?

Use a runtime secret store, minimum scopes, a test environment, and a documented rotation and revocation path.

Data and signals

Does a website visit prove buying intent?

No. It is a dated observation that may support a hypothesis after identity confidence, context, and exclusions are reviewed.

Is enrichment accuracy guaranteed?

No. Coverage, freshness, identity resolution, and field agreement vary by source and cohort.

Outreach controls

Can the workflow send automatically?

This site recommends keeping research, drafting, approval, and sending separate. A named person should approve relevance and applicable regional requirements.

What should a reviewer record?

Evidence, observation dates, inference, role rationale, verification, suppression status, opt-out handling, and the final decision.