02 / AN ACCOUNT RESEARCH LIBRARY

Know why
you’re writing.

DEMO RESULT1 account → 3 reviewed filesResearch, qualification, and a grounded outreach draft
Abstract evidence fragments converging into a qualified account signal before passing through a review gate.
WORKED TRANSFORMATION · FICTIONAL DATA

One account signal. One message with a reason to exist.

One supplied signal becomes research, qualification, and a review-ready draft.

INPUT · ACCOUNT EVIDENCE

Harbor Learning

The company publishes a weekly newsletter and is starting weekly workshops. Budget, internal bottlenecks, buyer authority, and purchase intent are unknown.

SYSTEM DECISION
The system marks the workshop launch as verified, treats workflow pain as an inference, and qualifies the account for a discovery question rather than a sales claim.
OUTPUT · TANGIBLE ASSET

Harbor Learning Account Pack

account-research.mdqualification.mdoutreach-draft.md
DECISION: Relevant for a discovery question.
NOT PROVEN: Need, budget, or intent.
DRAFT: Ask how workshop material reaches the newsletter.
Demo approval · revision 2 · source grounding 5/5 · risk handling 5/5

Fictional worked example. The download is a real ZIP produced by this demo. Approval, timing, and business results remain simulated.

GTM evidence ladder labeling a workshop launch as verified, workflow pain as inference, and buying intent as unknown.
Evidence ladderThe model earns specificity by showing what it does not know.
Three-part outreach message showing a verified signal, its relevance, and an honest question.
Message anatomyOne verified signal supports one relevant question. It does not prove intent.
SYSTEM MAP · EVIDENCE BEFORE ELOQUENCE

How evidence becomes a message worth reviewing

Evidence enters first. A human decides if the message leaves.

  1. INPUTAccount signal
  2. RESEARCHEvidence pack
  3. GATEHuman review
  4. OUTPUT3 files
Open technical architecture
  1. 01 · INPUTTarget accountsICP, trigger, offer, supplied public evidence
  2. 02 · RESEARCHEvidence packageSource excerpts, known facts, explicit unknowns
  3. 03 · ASSISTQualify + personalizeFit decision and grounded outreach draft
  4. 04 · STOREAccount revisionAssets, citations, risks, edits, history
  5. 05 · HUMAN GATEEvaluate + approveGrounding, fit, risk handling, do-not-contact check
  6. 06 · OUTPUTManual outreachReviewed ZIP, CRM-ready envelope, replies and outcomes
Trust boundaryThe public demo makes no network research or outbound send. Operator sources drive the draft.
Exception loopWeak evidence, invented intent, or do-not-contact status blocks release and returns the account for correction.
Evidence boundaryReplies, meetings, and revenue require verified receipts. Demo results never enter real metrics.
View ASCII architecture
TARGET ACCOUNTS + ICP
        │
        ▼
  SOURCE RESEARCH ── weak evidence ──► RESEARCH AGAIN
        │                                  │
        ▼                                  └──────┐
 QUALIFICATION + PERSONALIZATION                  │
        │                                         │
        ▼                                         │
 ACCOUNT REVISION + CITATIONS ◄───────────────────┘
        │
        ▼
 HUMAN REVIEW + DNC GATE ── reject ──► REVISE
        │ approve
        ▼
 OUTREACH ZIP → MANUAL SEND / CRM → REPLY + VALUE TRACKING
INTERACTIVE DEMO · Fictional examples only. No AI calls, sending, or publishing. Edits stay in this browser tab session. All results are simulated. Do not enter private data.
Load the demo or add a source package to begin.