# Product Brief — Conversational AI Reviews & Reputation Engine

> **Working name:** *ReviewLoop* (placeholder — alternatives: *StarLoop*, *Sabr*, *Rasa*). Pick later.
> **Status:** Direction locked 2026‑07‑01 after 3 research passes + a 5‑lens decision analysis.
> **One‑liner:** *Get more 5‑star Google reviews on autopilot — a WhatsApp‑native AI that has a real conversation with your customers, sends happy ones to Google and unhappy ones to you, in their own language.*

---

## The problem

Local businesses live or die by their Google rating (it drives Maps ranking → calls → revenue; a 1‑star rating increase is tied to ~5–9% more revenue). But **happy customers stay silent and unhappy ones shout** — so most businesses have fewer and worse reviews than they've earned. The old fix (SMS/email "please review us" blasts from Podium/Birdeye/etc.) gets low response, is English/US‑centric, has no intelligence, and costs $300–1,200/mo on annual contracts SMBs resent.

## Ideal customer (beachhead)

**UAE health‑&‑beauty clinics — dental, dermatology/aesthetics, med‑spas.** Reviews are existential to patient acquisition; owners are commercially sophisticated and pay $99–149/mo without blinking; churn is lower than restaurants/home‑services. **Channel: WhatsApp** (the default Gulf comms channel — Podium has *no* WhatsApp; Western tools are email/US‑SMS‑first). **Geography: Dubai + Abu Dhabi clinic clusters**, where the founder is physically present and can sell hand‑to‑hand.

> The **product is global‑capable**; only the **first ~30 customers are local**. Same engine travels to every WhatsApp‑heavy market later.

## What it does (the loop)

```
Patient visits → WhatsApp reaches out (right language) → AI has a short adaptive conversation
   → HAPPY: one-tap Google review link   |   UNHAPPY: private recovery + instant owner alert
   → every chat analyzed → owner sees rating climb + WHY, in a simple dashboard
```

## The wedge (why this isn't a me‑too)

1. **Conversational AI interviewer** — it *talks and probes* ("glad it went well — anything stand out?") instead of a dead form. Higher response + richer feedback. Incumbent survey tools structurally can't do this (genuinely new in 2025–26).
2. **WhatsApp‑native** — the right channel globally; a structural blind spot for the incumbents.
3. **Multilingual** — Arabic + English at launch, any language later.
4. **Insight extraction at SMB price** — themes/sentiment/"why behind the stars" that incumbents charge $1,000+/mo for.
5. **Compliant‑by‑design routing** — everyone is offered the public review (no illegal gating); unhappy customers additionally get fast private recovery.

## MVP v1 feature set

1. **WhatsApp Cloud API conversational review agent** — short adaptive 2–3 turn conversation after a visit (trigger: CSV / "send request" button in v1, webhook later).
2. **FTC/Google‑compliant routing** — always expose the public Google review link to everyone; LLM sentiment additionally opens a private recovery channel for unhappy customers + instant owner alert. *Compliance is an enforced guardrail, not an afterthought.*
3. **LLM sentiment + intent classification** on every conversation, with a **human‑in‑the‑loop** validation view.
4. **Insight extraction** — aspect‑level theme clustering + operational alerts in a simple dashboard.
5. **Google Business Profile API** — pull rating/review velocity; post AI‑drafted owner replies (one‑tap approve). *(Reviews are collected via a Google review **link**, not the API — the API can't post reviews on a customer's behalf.)*
6. **Multilingual** (Arabic + English, auto‑detect + translate).
7. **Self‑serve multi‑tenant onboarding + dashboard** — connect Google Business Profile, connect WhatsApp number, see star‑trend / velocity / themes / recovered‑customer alerts; weekly WhatsApp/email digest.
8. **Engineering backbone (the real learning):** per‑tenant data isolation, an eval harness with golden sets for sentiment/routing/theme extraction + compliance regression tests, tracing/observability for LLM cost/latency/quality.

## Monetization

Self‑serve SaaS, transparent, **no annual contract** (removing the friction SMBs resent). Per‑location tiers:

| Tier | ~Price/mo | Includes |
|---|---|---|
| Starter | $49 | Collect + compliant routing + AI owner replies |
| Growth | $99 | + multilingual + theme/sentiment insights |
| Multi | $149 | + multi‑location / agency white‑label |

**Target $5–10K MRR ≈ 50–100 clinics at $99–149** — at/above the median profitable micro‑SaaS (~$4.2K MRR), not a venture moonshot.

## Compliance & platform notes (verified)

- **No review gating** — selectively routing only happy customers to public reviews is prohibited (FTC + Google). Everyone is always offered the public review; the AI only *additionally* routes unhappy customers to private recovery. Ship this as a tested guardrail.
- **Google Business Profile API** — can **read** reviews (`reviews.list/get/batchGet`) and **reply** (`reviews.updateReply`); **cannot create** a review via API. Access is gated: Google approval + OAuth `business.manage` scope, GBP verified 60+ days, website required.
- **WhatsApp Cloud API** — per‑message pricing (since Jul 2025); **inbound replies within the 24h window are free**; marketing/auth templates are charged → design around inbound + utility.
- **UAE PDPL** — no vendor registration burden; honor opt‑out for direct marketing.

## Expansion (later, same engine)

Rings with near‑zero re‑engineering: UAE clinics → GCC clinics → other WhatsApp‑heavy markets (India, SEA, Nigeria, LatAm) → adjacent review‑dependent verticals (salons, wellness, restaurants). **Phase 2 = short‑let/holiday‑home operators** (they also live by reviews) → then 24/7 guest concierge + upsell, where the founder's **native‑iOS mastery** finally becomes a defensible moat.

## Riskiest assumption (validate FIRST, before heavy build)

**Durable willingness‑to‑pay / churn** — will clinics keep paying past the month‑one "review burst"? SMB SaaS churns 3–5%/mo. The recurring‑income thesis depends on continuously surfacing fresh review flow + insight value so it isn't a one‑time campaign.

**Cheapest validation steps:** (1) interview 15–20 reachable Dubai clinics on how they get reviews + what they'd pay; (2) **Wizard‑of‑Oz** — run the WhatsApp loop *by hand* for 3–5 clinics, measure response lift + net‑new reviews; (3) landing page + real pricing, measure trial conversion. See `go-to-market.md`.