Where we stand (v2.0, AI-speed revision). Private MVP shipped with hero photos, in-app help AI, operator dashboard, mobile/PWA optimization, login gate, four custom domains live (giftcue.app, giftcue.ai, giftcue.org, getgiftcue.com), and a public coming-soon page with waitlist email capture. Three rounds of QA, all critical issues patched. Zero external surface yet.
Board decisions taken May 11, 2026. V1.1 greenlit, four blocking decisions resolved, trademark deferred to $5-10K MRR, $300/mo vendor API budget approved, dual-side monetization confirmed. Hugo hired as CEO with operational authority; Jason in board/investor role with <1 hr/week time commitment. North star: prove an AI can run a profitable SBU end-to-end with minimal human time.
The actual product is good. AI quality holds across 12 personas (ages 14–86, budgets $25–$500+), images render in 2-3s at editorial quality, end-to-end flow works, mobile UX is right, PWA-installable, help AI knows context and refuses to fake actions. Operating model: ~$3.2M ARR Y3 with 85-87% gross margins, $15-25M exit target in late 2027/early 2028.
Compressed timeline (v2.0). V1.1 builds this week (May 12-17), not Q3. Friends-and-family beta next week. First public soft launch on getgiftcue.com end of May. Father's Day 2026 (June 21) is the first real public marketing push. $1K MRR target pulled forward 6 months. $10K MRR target pulled forward 9 months. Revenue start gain: ~$120K incremental Y1.
GiftCue is the productized evolution of the Mother's Day Picker. It solves a real, recurring problem: gift-givers who love someone but don't know what to get them, particularly when there's a knowledge gap between giver and receiver (adult kid → elderly parent, husband → wife, in-laws, long-distance relationships).
The product wedge is not novel tech — it's the combination of three things competitors don't put together: AI personalization tuned to the recipient (not the sender's guess), accessibility-first UX (not the modern consumer pattern that loses older recipients), and multi-occasion adaptive curation (not single-vertical). The MVP proves all three work.
| Surface | URL | Status |
|---|---|---|
| Landing | giftcue-may2026.jason-8ce.workers.dev | Live, internal |
| Sender form (build a picker) | /new | Live |
| Recipient picker | /g/:session_id | Live, AI-personalized |
| Recipient celebration | /g/:session_id after pick | Live, hero card + next-steps |
| Sender status (track pick) | /s/:session_id | Live, auto-refresh 30s |
| Operator console (admin) | /admin?token=… | Live, fulfillment tracking |
| AI help widget (Cue) | Every page, bottom-right "?" | Live, session-aware |
| PWA manifest | /manifest.webmanifest | Live, installable |
| Service worker | /sw.js | Live, caches hero images |
| Hero photos per option | /img/:sid/:idx | Live, Workers AI Flux |
| Business plan doc | giftcue-business-plan-may2026.jason-8ce.workers.dev | Published |
| QA report (round 1) | giftcue-qa-report-may2026.jason-8ce.workers.dev | Published |
Demo session anyone can try: Judith's Mother's Day picker (currently in celebration mode). Sender view: /s/7f5LhLJPYE7OqO0zk-nRzz87.
Wrote a four-path commercialization memo for the original Mother's Day Picker. Recommended killing consumer marketplace, building venue-facing SaaS. Jason then pivoted to the universal consumer gift picker concept — superseded the venue SaaS path, but the four-path scoring framework is preserved in the prior memo.
Single Cloudflare Worker. KV for sessions (30-day TTL). Service binding to existing mdp-api-proxy-may2026 for Anthropic AI calls (avoids same-zone subrequest recursion error 1042). Sender form, AI option generation (8 options per session), recipient picker (MDP UX pattern), pick recording, sender status. Full E2E verified.
Switched cards from text-only to vibrant editorial-quality. Added Workers AI binding (@cf/black-forest-labs/flux-1-schnell) for per-option hero image generation. AI prompt extended with photo_prompt and highlights fields. /img/:sid/:idx route with KV caching, 30-day immutable browser cache. Background image warm-up via ctx.waitUntil so images are ready before the recipient opens the link. Average generation: 2-3s per image. Verified across 4 personas; quality is genuinely editorial.
Replaced the empty post-pick screen with a server-rendered celebration view: hero photo, ribbon, category badge, price (only here, not on browse cards), description, "what people love" pull-quote, "why this matches", two next-step callouts ("Jason takes it from here", "What made this one stand out"), and "Want to look again?" reset button. Added consistent Back / Start over navigation across every page in the sequence.
Floating "?" button on every page. Opens a chat panel. Backend at /api/help proxies to Claude with a system prompt that knows GiftCue, knows the current page, knows the session (recipient name, sender name, occasion, budget, current pick), and is explicitly forbidden from inventing features or promising follow-ups GiftCue can't deliver. Page-specific quick-suggestion chips. Conversation history preserved client-side (last 6 turns sent to AI). Verified: correctly explains the product to a confused recipient, admits "no automatic SMS" to a sender, refuses off-topic requests politely.
Safe-area insets for notched iPhones, 100dvh instead of 100vh, font-size: 16px on all inputs to prevent iOS auto-zoom, help panel becomes bottom sheet under 640px width, viewport-fit=cover, theme-color, apple-mobile-web-app-capable, PWA manifest with installable display mode, minimal service worker (caches hero images), SVG icon. Operator console at /admin?token=… shows all sessions with fulfillment status, "Mark fulfilled" button, total/opened/picked/fulfilled counters, and per-session sender + recipient links. Event log in KV under evt:<ts>:<id> keys captures session creation, picks, and fulfillment marks.
Round 1 (5 passes): functional/AI quality across 8 personas, security/safety, accessibility (WCAG 2.1 AA), performance/cost, failure modes. Found and fixed: input length caps, keyboard chip accessibility, color contrast, focus-visible, reduced-motion, skip-link, aria-live. Round 2 (5 passes): photo quality across 4 personas, image caching/fallback, visual rendering, sender preview cards, background warming. Round 3 (2 passes): mobile-specific tests, PWA validity, admin/event log, regression across all flows. Zero blocking issues remaining; all critical patches deployed in-session.
┌─────────────────────────────────┐
│ Recipient browser │
│ (mobile-first, PWA-installable)│
└────────────┬────────────────────┘
│ HTTPS
▼
┌──────────────────────────────────────────────────────────┐
│ giftcue-may2026 (Cloudflare Worker, ~50 KB) │
│ Routes: │
│ GET / landing │
│ GET /new sender form │
│ POST /api/sessions create session + warm imgs │
│ GET /g/:sid picker OR celebration (by state) │
│ GET /s/:sid sender status │
│ GET /img/:sid/:i AI-generated hero (KV-cached) │
│ POST /api/help Cue (session-aware AI help) │
│ GET /admin operator console (token-auth) │
│ POST /admin/fulfill mark gift as delivered │
│ GET /manifest.webmanifest, /sw.js, /icon.svg │
└────┬──────────────┬─────────────────┬──────────────────┬─┘
│ │ │ │
▼ ▼ ▼ ▼
┌─────────┐ ┌────────────────┐ ┌──────────┐ ┌──────────────┐
│ KV │ │ AI_PROXY │ │ AI │ │ Workers AI │
│ giftcue-│ │ (service │ │ binding │ │ Flux Schnell │
│ sessions│ │ binding to │ │ (image │ │ (image gen) │
│ │ │ mdp-api-proxy) │ │ gen) │ │ │
│ 30d TTL │ │ │ │ │ │ │
└─────────┘ └────────┬───────┘ └────┬─────┘ └──────────────┘
│ │
▼ ▼
┌──────────────────────────────┐
│ Anthropic API │
│ (option generation + Cue help)│
└──────────────────────────────┘
giftcue-may2026 → mdp-api-proxy-may2026.worker.js file, deployed via wrangler CLI. Fast iteration.| Resource | Identifier | Cost |
|---|---|---|
| Worker | giftcue-may2026 (~50 KB, 38 KB gzipped) | Free up to 100K req/day; $5/mo above |
| Worker (AI proxy) | mdp-api-proxy-may2026 | Same plan, ~$0 |
| KV namespace | giftcue-sessions (id 67d44039f73d4cf0a1458aaa0ee82083) | Free up to 100K reads/day, 1K writes/day |
| Workers AI binding | AI | Free up to 10K neurons/day (~1K images); $0.0125/1K above |
| Anthropic API | Via mdp-api-proxy | Pay-per-token (~$0.02 per option-set, ~$0.005 per help reply) |
| Custom domain | None yet (jason-8ce.workers.dev) | $10–15/yr when bought |
| SMS provider | None (V1.1) | Twilio ~$0.01/SMS |
| Payment processor | None (V1.1) | Stripe 2.9% + $0.30/txn |
| Curation API | None (V1.1) | Yelp Fusion free 5K/day, Google Places $$, Tremendous % |
| Volume | Monthly cost | Cost / picker | Notes |
|---|---|---|---|
| 100 pickers/day | ~$15 | $0.005 | AI text + 8 images dominate |
| 1,000 pickers/day | ~$150 | $0.005 | Mostly AI |
| 10,000 pickers/day | ~$1,500 | $0.005 | Swap to Claude Haiku to cut text AI cost 10x |
| 100,000 pickers/day | ~$15,000 | $0.005 | Migrate KV → D1 + Postgres; consider image CDN |
Numbers assume Haiku for text gen post-V1, Flux Schnell for images, KV for hot data. Lower than the prior estimate because the Haiku swap is a guaranteed cost reducer once tested.
Per your direction: the plan is to start as a web app, then ship native iOS and Android apps to the official stores. Three sequential phases. PWA today (done), Capacitor wrap to both stores (the next big milestone), then native rewrite only if scale forces it.
viewport-fit=cover + safe-area insets (notched iPhones don't crop)100dvh instead of 100vh (handles iOS Safari URL-bar shifting)font-size: 16px (prevents iOS auto-zoom on focus)display: standaloneIonic Capacitor takes the existing web app and produces native iOS (Swift) and Android (Kotlin) shells around it. Same web codebase. Native APIs available (push notifications, biometric auth, share sheet, contacts, camera). Distributed through the official stores. This is the cheapest path to "real apps" — about 1-2 weeks of focused work versus 3-6 months for a native rewrite.
| Item | Apple | One-time / annual | |
|---|---|---|---|
| Developer account | Apple Developer Program | Google Play Console | $99/yr · $25 one-time |
| Mac required for build? | Yes (Xcode) | No (Android Studio on any OS) | — |
| App signing | Apple-managed certs + provisioning profiles | Google-managed via Play App Signing | — |
| Store listing | App name, subtitle, description, keywords, 5–10 screenshots per device class, app icon (1024×1024), privacy policy URL, support URL | App name, short + long description, 2–8 screenshots, feature graphic, app icon (512×512), privacy policy URL | Prep ~1 day each |
| Review time | 1–3 days typical; can spike to 2 weeks at holidays | 2–24 hours typical | — |
| App review pitfalls | Apple rejects PWA-wrapped apps that don't feel "native enough" or lack offline capability. Must use native UI patterns (share sheet, system fonts). No web-only content. | Much more lenient. Mostly checks for spam, deceptive practices, content policy. | — |
| Payment for in-app purchases | 30% Apple cut (15% under $1M/yr Small Business Program) | 30% Google cut (15% on first $1M/yr) | OR use Stripe for non-digital fulfillment (gift card purchases are physical goods, exempt) |
npm init @capacitor/app, copy current web codebase, configure capacitor.config.json with bundle ID (com.giftcue.app), install plugins (push notifications, share, browser, status bar).@capacitor/assets CLI from a 1024×1024 source image.| Line item | Cost |
|---|---|
| Apple Developer Program (annual) | $99/yr |
| Google Play Console (one-time) | $25 |
| Firebase (push notifications via FCM) | Free at our volume |
| APNs cert | Included in Apple Developer |
| App icon + screenshot design | $0–500 (DIY in Figma or hire on Fiverr) |
| Build time | ~60 hrs (1.5 weeks focused) |
| Ongoing maintenance | Same web codebase — just rebuild + resubmit when shipping web updates |
| Total to launch in both stores | ~$125 + ~60 hrs |
| Option | Pros | Cons | Verdict |
|---|---|---|---|
| Capacitor | Same web code, native shell, ~1-2 weeks to ship both stores, official Ionic Foundation backing, used by Burger King, Sworkit, Untappd | WebView performance ceiling (fine for our use case) | Pick |
| React Native | True native UI, large ecosystem | Full rewrite (~3 months), JavaScript bridge complexity, team specialization needed | Skip until 50K+ DAU |
| Flutter | Single codebase, beautiful default UI | Dart language; full rewrite; no web codebase reuse | Skip |
| Swift + Kotlin native | Best performance, full platform access | Two codebases, two teams, ~6 months | Only at major scale |
| PWA-only forever | Zero additional cost | No Apple App Store presence (Apple deliberately limits PWAs), discovery problem, no push on iOS until very recently | Insufficient for the stated plan |
If we hit 50K+ DAU and the WebView ceiling becomes a real constraint (rare for our use case), consider a React Native or fully native rewrite. Cost: $50K-200K + ongoing engineering. Should be funded by a real revenue base, not bootstrapped.
| Milestone | Phase | When |
|---|---|---|
| PWA available on web | 1 | Now (done) |
| V1.1 web launch (real curation, payments, SMS) | 1 | Aug–Oct 2026 |
| First 100 paying senders + push-notification demand confirmed | — | Dec 2026 |
| Capacitor wrap + App Store + Play Store submission | 2 | Jan–Feb 2027 |
| Both apps live in stores before Mother's Day push | 2 | Apr 2027 |
| Native rewrite (if needed) | 3 | 2028+ |
Build through Phase 1 (PWA web app) to V1.1 launch. Once we have any meaningful paying-user base and confirmed demand for push notifications, ship the Capacitor wrap to both stores within 2 weeks. Don't rewrite native unless scale forces it. Total cost to be in the App Store + Play Store: ~$125 plus 1.5 weeks of work. The web codebase is the same one shipping in the stores — every web update is also an app update.
You flagged: this app is getting valuable; lock it down. Done. The main GiftCue worker is now behind a password gate while recipient links stay public (they have to — recipients have only the link).
/ — landing/new — sender form/s/:sid — sender status page/admin — operator console (token check now layered on top of login)POST /api/sessions — session creation/g/:sid — recipient picker / celebration/img/:sid/:idx — hero imagesPOST /api/sessions/:sid/pick — recipient's pickGET /api/sessions/:sid — recipient status checkPOST /api/help — Cue help widget (recipients use this)/manifest.webmanifest, /sw.js, /icon.svg — PWA assets/login, /logout — auth endpoints themselves/login with a single password. Default: BT-GiftCue-2026-PrivateBuild!giftcue_auth with 14-day expiry/login?return_to=<original>cd /tmp/giftcue-deploy wrangler secret put LOGIN_PASSWORD # Paste new password when prompted, hit enter wrangler secret put AUTH_SECRET # Paste a new long random string (e.g., openssl rand -hex 32) # Rotating this invalidates all existing sessions
The current gate is fine for an internal build. For V1.1, migrate to Cloudflare Access (Zero Trust) for SSO, multi-factor auth, audit logs, and IP allow-listing:
giftcue-may2026.jason-8ce.workers.dev (or the custom domain when bought)/g/*, /img/*, etc.)Until that's done, the current cookie-based gate is doing the job.
You asked: where do notifications go when a recipient picks? Honest answer: today, only to your sender status page (which auto-refreshes every 30s). If you're not on that tab, you don't know.
To fix that, this session added:
session_created, picked, and fulfilled event is logged under evt:<ISO timestamp>:<id> keys with 1-year TTL./admin?token=… — shows total / opened / picked / fulfilled counts, every session with current state, "Mark fulfilled" button, links to sender + recipient views.This is the V0 of fulfillment tracking. It's enough for Jason to handle 50–500 gifts/month manually. Beyond that, real notifications (email/SMS to your phone when a pick happens) and a real queryable database are required.
| Stage | Volume | Storage | Notifications |
|---|---|---|---|
| V0 (now) | 0–500 gifts/mo | KV event log + admin page | Visit /admin to check |
| V1.1 | 500–5K/mo | KV + nightly D1 sync | Email-to-self via Resend when a pick happens; SMS to sender via Twilio when their recipient picks |
| V2 | 5K–50K/mo | D1 (Cloudflare) or Neon Postgres | Push notifications via PWA (web push) + email digest + Slack webhook for operator |
| V3 | 50K+/mo | Postgres + Redis cache; analytics warehouse (BigQuery) | Real-time push, ops dashboard, automated fulfillment routing |
What this costs to build at each stage: V1.1 ≈ 4 hours of work (Resend $0/mo free tier, Twilio ~$0.01/SMS). V2 ≈ 2 weeks (D1 migration, push notification certs, Slack integration). V3 = a real engineering hire or contract — outside the scope of bootstrap mode.
The picker is only valuable if the gift actually arrives. Five tiers, from cheapest to most complex:
Recipient picks. Sender sees the pick on /s/:sid. Sender clicks the venue's phone number to buy a gift card, or visits the website. Sender marks "fulfilled" in admin. Margin: 100% on the $7 picker fee.
One-click "buy this gift card" button after the recipient picks. Integrate Tremendous or Giftbit API — they support 1,000+ merchants with REST API. Sender pays $50 + $7 picker fee + 4% processing. We collect a $2-4 spread per gift card. Margin per gift: $9-11.
For local experiences picked through GiftCue: integrate Yelp gift card API (free tier), Resy reservations, OpenTable. Recipient books directly through the app. We collect a referral commission (5-15% from the venue) plus the picker fee. Margin per gift: $15-25.
Premium tier ($29/mo or $199/yr). A human (or an AI agent — V3) handles the entire fulfillment: makes the booking, wraps the physical gift, sends the recipient a custom note from the sender, follows up with both parties. Margin: ~60-70% (humans are the cost).
"Mystery monthly gift" for a recipient. Sender pays $29/mo subscription. Each month, GiftCue selects a new gift, fulfills it, sends. Recipient gets a surprise. Sender stays out of it. This is the highest-LTV product variant.
You asked: ways to make money on both ends — sender and vendor. Here's the full take.
| Stream | Price | Margin | Notes |
|---|---|---|---|
| Pay-per-picker | $7 / gift | ~95% | One-time use, no commitment |
| Personal subscription | $9 / mo or $79 / yr | ~87% | Unlimited pickers, recipient memory, occasion reminders |
| Family subscription | $14 / mo or $129 / yr | ~85% | 4 sub-accounts, shared recipient profiles |
| Concierge tier | $29 / mo or $199 / yr | ~60% | Human fulfillment, custom notes |
| Mystery monthly | $29 / mo per recipient | ~50% | Recurring surprise gift; highest LTV |
| Gift-of-GiftCue | $49 (3 mo) or $149 (12 mo) | ~70% | A subscription as a gift |
| Stream | Price | Notes |
|---|---|---|
| Affiliate commission | 4-15% on gift card / booking value | Default with Amazon, Tremendous, Yelp, Cloud9. No vendor relationship needed. |
| Featured placement | $50-200 / mo per category in a metro | "Top spa pick in Asheville this Mother's Day." Auctionable, sortable, capped. |
| Verified vendor listing | $29 / mo | Vendor pays to be in the curated set; gets a profile, photo, real reviews integrated. |
| Per-redemption fee | $3-10 / redemption | Vendor pays only when their gift card is actually used. |
| Booking commission | 10-25% on experience bookings | For experiences booked through the platform. |
| Aggregate demand data | $500-5K / month per vertical | Sell anonymized "what people are gifting in your category" data to spas, wellness venues, gift card networks. |
| White-label license | $2-10K / month per partner | Senior-care orgs, hospice, chambers of commerce license the picker tech. |
White-label and aggregate-data are V2 plays once you have a year of pick data to sell.
v1.0 assumed a human consultant operator working 8 hrs/week with V1.1 launch in Q4 2026. v2.0 reflects AI-speed compression: V1.1 ships this week, soft launch end of May, Father's Day 2026 as first real public push. Y1 expands from a 4-month active revenue window to a 12-month one. Revenue compounds earlier, not bigger overnight — customer acquisition still happens at human pace, but we get an extra 6-8 months of compounding.
| Metric | Y1 (May 2026 – Apr 2027) | Y2 (May 2027 – Apr 2028) | Y3 (May 2028 – Apr 2029) |
|---|---|---|---|
| Active paying senders (exit-of-year) | 2,400 | 12,000 | 45,000 |
| Average gifts/sender/yr | 3.5 | 5 | 6 |
| Gift transaction revenue ($7 picker) | $58K | $320K | $1.55M |
| Subscription revenue ($9/mo blended) | $32K | $240K | $820K |
| Affiliate / commerce | $22K | $160K | $500K |
| Featured / vendor placement | $8K | $60K | $240K |
| White-label / API | $0 | $70K | $220K |
| Total revenue | $120K | $850K | $3.33M |
| Operating costs | $35K | $185K | $520K |
| Net contribution | $85K | $665K | $2.81M |
| Net margin | 71% | 78% | 84% |
| Exit ARR (run-rate at year-end) | ~$280K | ~$1.6M | ~$5M |
Operating cost breakdown (Y2): AI ($35K — Sonnet text + Flux images at scale, Haiku swap saves 70%), Stripe processing ($28K), CF infra ($8K), paid marketing ($85K spread holiday-heavy), Resend + Twilio + Yelp + Tremendous APIs ($14K), 1 part-time offshore CS / fulfillment hire ($15K).
Y1 revenue by quarter (v2.0):
| Quarter | Revenue | Driver |
|---|---|---|
| Q2 2026 (May-Jun) | $2K | Beta + first Father's Day test pulse |
| Q3 2026 (Jul-Sep) | $10K | SEO compounding, summer birthdays, content + organic |
| Q4 2026 (Oct-Dec) | $35K | First Christmas push, scaled content arsenal |
| Q1 2027 (Jan-Mar) | $30K | V-Day push, post-holiday retention |
| Q2 2027 (Apr-Jun) Mother's Day | $43K | Year-defining push, app stores live |
| Y1 total | $120K | vs. prior $29K plan — ~4× compression gain |
Exit math (revised): Y3 ARR ~$5M with 84% net margin. Comparable consumer-software/marketplace transactions trade at 4-8x ARR. Target acquirer set unchanged: Tremendous, Goody, Stamps.com, Constant Contact, Caring.com, A Place For Mom. Realistic exit: $20-40M in late 2028 / early 2029. Stretch case (true viral product/market fit, 50K+ DAU): $60-120M.
Hugo's note as CEO: Y1 of $120K is conservative. The aggressive case (if Father's Day 2026 converts well + Christmas content arsenal lands + we land one viral TikTok in Q4) is $200-280K Y1. I'll forecast the conservative case to the board and execute toward the aggressive.
| Milestone | Target date | Trigger / Gating |
|---|---|---|
| V1.1 build complete (Yelp Fusion, Stripe, Twilio, Resend, magic-link, rate limit, Haiku swap) | Fri May 15, 2026 | Hugo executes; no external gate |
| Stripe + Twilio account verification cleared | ~May 20, 2026 | Real-world clock (KYC, A2P 10DLC) |
| Friends-and-family beta (10-20 senders) | Wed May 20, 2026 | Jason sends to 10 contacts |
| Soft launch — getgiftcue.com goes from coming-soon to active | Mon May 25, 2026 | Beta data shows pick-rate ≥40% |
| First Father's Day push | Sun Jun 21, 2026 | First real paid + content test |
| First $1K MRR | Late Jun 2026 | ~150 paying senders (was Dec 2026) |
| First $10K MRR (cash-flow positive) | Sep 2026 | ~1,200 senders cumulative (was Jun 2027) |
| First Christmas push (scaled) | Dec 2026 | Content arsenal of 40+ SEO posts indexed |
| Capacitor wrap → Apple App Store + Google Play | Jan-Feb 2027 | Wait for store reviews (real-world clock) |
| Mother's Day 2027 push (year-defining) | Sun May 9, 2027 | Native apps live + scaled marketing |
| $50K MRR (CEO salary covered + reinvestment runway) | Jun 2027 | ~4K paying senders (was Dec 2027) |
| $100K MRR ($1.2M ARR — exit-ready) | Q4 2027 | Pulled forward 6 months from v1.0 |
| Exit window opens | Q1-Q2 2028 | $3M+ ARR run-rate, growth still trending up |
Compression analysis:
Profitability: Gross profitable from week 1 (~85% margin, no COGS). Net profitable when paid marketing in Q4 holiday push pays back inside its cohort — projected: November 2026 onwards.
| Risk | Severity | Mitigation |
|---|---|---|
| Consumer CAC is too high to justify $7-9 pricing | HIGH | Friends-and-family beta validates LTV before any paid spend. Kill if CAC > LTV after 90 days of paid testing. |
| Sender doesn't know recipient well enough → bad picks → no repeat | HIGH | Form prompts extract specifics. AI handles vagueness. Add "low-info recipient" branch in V1.1 where AI asks 2 follow-up questions before generating. |
| Recipient privacy issues when real SMS goes live | HIGH | Sender attests permission. Recipient SMS includes opt-out from message one. Honor STOP. Never collect recipient PII beyond name + city. |
| AI hallucinates non-existent local venues | MED | V1.1 swaps AI candidate generation to real Yelp/Places results, AI just curates. |
| Distraction from BOSSTORQUE consulting revenue | MED | Hard cap: 8 hrs/week on GiftCue. Build runs autonomously where possible. Pause during BT client crunch. |
| Goody or another player launches the same wedge | MED | Goody tried consumer and pivoted to B2B — informative. Our wedge is sharper (accessibility-first + AI-personalized + multi-occasion). Speed-to-market matters; aim for pre-Mother's-Day 2027. |
| Brand confusion with BOSSTORQUE | MED | Separate brand, separate domain, no BT references on public site. |
| Sunk-cost bias — keep building because we built | HIGH | Pre-commit to the V1.1 decision gate. If friends-and-family NPS <40, archive. |
| Workers AI rate-limit / outage | MED | Background warm-up + KV cache means single-image outages don't break a session. Add a real image fallback CDN (Unsplash) as backup. |
| Fulfillment fails at scale (gift cards don't arrive) | MED | Tier 2 fulfillment uses APIs with SLAs (Tremendous 99.9%). Tier 4 concierge has human override. |
| Decision | Recommendation | Cost of yes | Cost of no |
|---|---|---|---|
| 1. Greenlight V1.1 | YES if 5/5 friends say "I'd use this for a real gift" | ~$300/mo ongoing + 8 hrs/wk of your time for 60 days | $0 — kill and archive |
| 2. Real name + domain | Run trademark check on GiftCue / GivePath / Picksy / EaseBook. Buy the cleanest. | $20 domain + $250-2K trademark | Stuck on jason-8ce.workers.dev forever |
| 3. Approve $300/mo vendor API budget | YES — Tremendous + Yelp Fusion + Stripe + Twilio + Resend together. Effectively free at our volume initially. | $300/mo capped | No real curation, no real payments |
| 4. Dual monetization model | YES — start sender-pay ($7 + $9/mo), layer in affiliate commission immediately, add featured placement in month 6 | Tracking complexity | Single-side revenue caps the upside |
Running this as your CEO, here's what I'd advise. — Hugo
1. Treat this as a real, separate business unit. Not a side project of BOSSTORQUE. Separate P&L, separate name, separate domain, separate brand. Yes, share the AI infra and learnings. But don't let GiftCue dilute BT's positioning to construction/field service clients.
2. Pre-commit to the kill criterion now. If 5 friends-and-family beta users in July 2026 don't independently tell you "this is something I'd actually pay for," shut it down. Don't let 6 months of work create a sunk-cost case for keeping it alive. The cost of running a failing consumer app is real: it eats focus, it eats reputation, it eats time.
3. Set up weekly cadence starting now. Every Monday: 15 minutes reviewing the operator console. Total sessions, total picks, fulfillment rate, any open questions in /api/help. Use this as the heartbeat. If the heartbeat goes flat for 4 weeks, escalate or pause.
4. Budget the next phase explicitly. V1.1 build = $300/mo infra + ~60 hrs of your time + ~$1K legal/domain/trademark. Total Q3 spend ≈ $2K + 60 hrs. Treat that as a budget. Anything beyond it requires a re-decision.
5. Document the IP value as you go. The accessibility-first conversational commerce UX pattern is the actual sellable thing. Patent? Probably not worth it. Trademark? Yes. But the most valuable asset is the play-by-play of what you built and why — that's what an acquirer pays for. Keep the build reports, the QA reports, this Megillah — they're the artifact set that makes a buyer trust the IP.
6. Set a maximum life of 24 months. If we're not at $50K MRR by May 2028, sell what's salvageable (the codebase, the trademark, the email list) to the highest bidder among the comparable acquirers, take the lesson, and move on. Don't run this past its prime.
7. The honest framing. This is a real, viable consumer SaaS bet with a sharper-than-average wedge, low-cost infrastructure, and a clear path to a $10-20M exit. It's also OUTSIDE your stated BOSSTORQUE thesis. Both things are true. The decision isn't whether GiftCue is good — it is. The decision is whether you want a second business and what you're willing to give up in BT focus to grow it.
The four blocking decisions are resolved (May 11, 2026). Hugo is operating. Board reviews weekly. North star: AI-led SBU to $3M+ ARR by FY29 with minimal human time.