Decisions log — newest first
Business — core · biz/DECISIONS.md

Decisions log — newest first

Only entries in this log are settled. Anything else in BUSINESS-MODEL.md is a recommendation until it appears here.

Stage note (Daniel, 2026-07-21). It's very early — we don't fully know what we want yet. The real work right now is experimenting: getting customers and testing the business model in the market. So keep this log light — few hard commitments, and treat even "settled" entries as provisional, to be revisited as we learn. Prefer testing things in practice over deciding them on paper.

2026-07-17 (late) — WhatsApp re-cut RATIFIED by Daniel + next steps settled

  • Ratified as settled: everything in the verdict below — WhatsApp Receptionist flagship, R1,800/mo capped tier, Twilio PAYG, post-Oct-2026 pricing basis.
  • PM-software research leg green-lit and LANDED (research/2026-07-17-sa-practice-tools.md): pitch splits by sub-vertical — dentists = cost-displacement (receptionist R12–21k/mo; GoodX has the only proven writable diary API), psychologists = after-hours capture (often nothing to displace; Google/Outlook Calendar sync = MVP integration, also the demo-build directive). Scope Bookem before the first dental pitch; test R1,800/mo willingness-to-pay on psychologist prospects (Alexa is the segment).
  • Live WhatsApp Receptionist demo = NEXT BUILD TASK (fictional practice, real number anyone can message): sales asset + site AI-demo upgrade + Q1 substrate dry-run in one. Scope in a dedicated session.

2026-07-17 (late) — WhatsApp research verdict

Hypothesis researched same evening (research/2026-07-17-whatsapp-sa.md): holds. Recommended re-cut (BUSINESS-MODEL.md addendum): WhatsApp = the SA chat surface; flagship = "WhatsApp Receptionist" for health/professional practices (genuinely underserved: aggregators above, R299–2k/mo generic SaaS below, R180k+ bespoke ceiling); SA chat retainer R1,500 → R1,800/mo with a 600-conversation cap + overage, transactional scope only, broadcasts priced separately; lean BSP (Twilio PAYG), never WATI-style bundles. Two clocks: Oct 1 2026 Meta meters the service window (price post-Oct from day one) and Aug 1 2026 Meta's native Business Agent bills (pitch = what native can't do: real calendar/PM integration, SA payments, POPIA, CRM). Onboarding realities: 2–14-day Meta verification, dedicated number, payment links not in-chat pay. Open: what PM/scheduling software SA practices actually run; volume expectations per client type. Reference engine: case-study one-pagers, one named founding client per package, live WhatsApp demo number, audits feed the proof library.

2026-07-17 — founding session (research + strategy panel + Daniel's calls)

Settled:

  • Frame: side income first — R20–50k/mo target within 12 months, day job stays, 10–20 h/week total. Honest plan band: ~R25–35k/mo net sustainable by m12; R40–50k = peak months via the intl price lever, not volume.
  • Markets: SA (ignition, ZAR ballast) + UK/EU/US (margin layer, ramps Q3).
  • ⛔ Hospitality fully excluded (employer separation). Restaurant guest app = anonymized capability proof only; never a referral seed.
  • Backbone locked: "Repaint" web rebuilds lead (R28k SA / $2,750 intl); AI chat agents are the recurring attach (R15k + R1,500/mo); automations opportunistic; mobile = wrapper on request. Full ladder in BUSINESS-MODEL.md.
  • Voice = P1/P2, not P0 (Daniel: the tech transfer from his existing voice work is cheap, so it's a committed fast-follow, not speculative — but it still waits for the metering/monitoring substrate and paying demand before it's SOLD).
  • Agency white-label = dual-rate: 10% finder's fee (Daniel keeps the client) or 25–30% wholesale (agency resells under their brand); Daniel always keeps the hosting/retainer layer. The 25–30% figure is unverified — validate with partner #1.
  • Intl Repaint anchor: $2,750 for the first 3–5 international clients, then raise toward $3,500–4,500 once references/reviews exist.
  • Channels sequenced: referrals (Alexa + Aerias asks this month) → agency partnerships Q2 → Upwork Q3. Cold email and BNI dropped on the numbers.
  • Client names cleared: Berlein Psychology, Aerias, Nsuna — public use OK.
  • Domain email: daniel@danielslater.dev this week (steps in ~/dan-hub/COMMANDS.md ▶3).
  • Stack: self-host from day one, reusing Daniel's OWN repos/patterns.
  • Tutoring: out of scope (route already redirects home).

Open / in flight:

  • WhatsApp as a major product (Daniel's hypothesis: WhatsApp ordering/booking/AI agents are a natural SA hit) — research running 2026-07-17 evening; lands in research/, then the chat-agent offer gets re-cut around it if it holds.
  • References & proof gap (Daniel: "wish we had more references") — needs a deliberate reference engine: case-study one-pagers for berlein/aerias, possibly 1 discounted beta client per package traded for a named case study, self-built live demos as proof.
  • Restaurant-app IP provenance — confirm it's clean of employer time/assets before any portfolio use; get written OK before reusing any Nsuna-side code.
  • SA-specific AI/automation pricing — no SA data existed in the sweep; chat/voice/ automation SA prices are set off the web-dev hourly proxy (WhatsApp research partially covers this).
  • Site funnel/pricing copy to inherit all of the above (separate design session).