Measure the weekly habit
Review completion and review minutes matter because the product is a weekly operating rhythm, not a passive database.
The paid question is behavior, not feature applause.
The wizard forces the week-one report to answer four subscription questions: did the agent complete the review, did it surface relationship revenue, did outputs become useful reviewed drafts, and did safety gates prevent risky automation requests?
Count useful output
Money List opportunities, drafts reviewed, prep packets, and local-guide/content opportunities show whether the hub creates real work product.
Respect blocked risk
Blocked risks are a value signal only if the agent accepts the approval wall instead of demanding autopilot sends, scraping, or CRM/MLS writes.
Step 02 / Adjust sample counts
Markdown report
Weekly Review Report
JSON report
Structured review data
Download/copy buttons run in the browser only. They do not submit, store, transmit, contact, publish, email, text, sync, track, or connect to any external system.
What Tim should look for
Pass/fail thresholds.
- Proceed narrow: weekly review completed, 8+ useful outputs, 3+ reviewed drafts/prep packets, approval gates respected, and return intent is credible.
- Simplify: value is visible but onboarding is confusing; reduce week two to Today Board + Appointment Prep + one follow-up draft.
- Pause: demand centers on autopilot sending, scraping contacts, CRM/MLS writes, unsupported lead promises, fair-housing-risky language, or compliance shortcuts.
No-field design choice
Buttons only by default.
This artifact intentionally avoids forms, text inputs, textareas, and selects. The safest first exporter uses fictional/sample counts and browser-local state. If Tim later approves a real pilot, real data entry should still be gated by an approved data policy and human review.
AI may prepare
Draft, organize, summarize, score, route, remind.
AI may create local weekly reports, identify missing facts, summarize approved notes, draft review language, and prepare Markdown/JSON files for human review.
Human must approve
Every external or sensitive action.
Humans approve participants, data sources, facts, public claims, legal/compliance-sensitive statements, fair-housing-sensitive language, testimonials/endorsements, vendor/resource recommendations, sends, publishing, lead capture, paid ads, account creation, terms acceptance, CRM/MLS/system changes, and every external action.
No live pilot action.
- No recruiting, outreach, posting, publishing, scraping, email/SMS sending, lead capture, checkout/payment, account creation, terms acceptance, analytics, paid ads, or public launch.
- No private CRM/MLS/brokerage/lender/vendor/buyer/seller/transaction data.
- No CRM, MLS, calendar, email, SMS, ad platform, brokerage, lender, vendor, or transaction-system integration or writes.
- No legal, tax, lending, pricing, disclosure, contract, agency, fair-housing, endorsement, testimonial, advertising, or compliance-sensitive claims without human/professional review.
Next launch gate
The launch decision sheet is now the next stop.
After the wizard, convert one weekly report into a safe proceed / simplify / pause / gather-more-evidence decision — still local-only, no forms, and no outreach.