# AI Assistance — competitor features to add

**Updated:** 8 Sep 2026 PT · **Ticket:** TM485 · **SKU:** Apex Receptionist $199/mo after-hours (Bay $249; CoS city test $99)  
**Rules:** Never Fuller. AUTHORIZE SEND NO. Not a human call center clone — AI voice + one shared book.

Public hub: https://apex.socialtokens.site/projects/ai-assistance/  
This page: https://apex.socialtokens.site/projects/ai-assistance/research/competitor-features.md

---

## Competitor map (2026)

| Competitor | Model | Sticker (approx) | Billing unit | What they sell well |
|---|---|---|---|---|
| **Ruby** | Live human US | ~$250–$720+ | Per minute (~50–200 min packs) | Warm human voice, brand trust |
| **Smith.ai** | Hybrid AI + human | AI from ~$0–low; human ~$300/30 calls | Per call (+ add-ons) | Law/pro services intake, CRM, spam free |
| **AnswerConnect** | Live human 24/7 | ~$325–$395+ | Per minute | Bilingual, always-on humans |
| **receptionist.ai / AI-receptionist.com** | AI-native | ~$14–$199+ | Flat + minute packs | Flat rate, calendar, spam filter, boss mode |
| **Virtual Reception AI** | AI-native | ~$41–$249 | Minute packs | Booking + SMS alerts + spam block |
| **Rosie / Upfirst / AIRA** | AI budget | ~$25–$160 | Call or minute packs | Cheap after-hours coverage |

**Our wedge:** flat **$199** for owner-named shops that **close** — answer after hours, qualify, **book next open**, **text the owner**. One extra completed job pays the month. Not per-minute surprise bills. Not Fuller. Not Jobber/HCP/ServiceTitan robot if already installed.

---

## Feature gaps — what to add (priority)

### P0 — ship / demo must prove (closes shops)

1. **Spam / robocall filter** — Ruby bills minutes on junk; Smith markets “no charge for spam.” Demo + product: drop obvious robocalls before they burn owner attention.
2. **Owner SMS/email morning recap** — competitors all push notifications. We already pitch “text the shop”; make a **one-screen morning list** (3 jobs, windows, caller names) part of every demo vertical.
3. **Calendar book that survives refresh** — demo slot book is localStorage; production needs Google Calendar (or shop’s book) so double-book fear dies.
4. **Warm transfer / “urgent only” escalate** — Smith/Ruby escalate humans. We stay AI-first: escalate **only** emergency keywords to owner’s cell (never Fuller), with transcript.

### P1 — differentiates vs cheap AI ($14–$49)

5. **Vertical-tuned intake scripts** — HVAC/plumbing/auto/dental/legal question trees (already partially in demos). Competitors are generic; we win on **owner-named trade language**.
6. **Shared one-book collision hold** — two callers cannot take the same slot (Demo A gate). Sell this hard vs voicemail and vs dumb AI that double-books.
7. **Personalized demo links** — `?business=NAME&owner=FIRST` already exist; keep three doors (voice / text / web) in every Track A draft.
8. **Call recording + transcript to owner** — table stakes for AI plans; include in $199 without add-on nickel-and-diming (Smith charges add-ons).

### P2 — later (after first paying shops)

9. **Spanish line** — Ruby includes bilingual; Smith add-on. Add when CoS opens bilingual metros.
10. **Payment collect on call** — Smith/Ruby add-on. Nice for restaurants/deposits; not required for trades MVP.
11. **CRM webhook (Jobber/Housecall login sync)** — only if shop asks; kill pitch if their AI receptionist is already on.
12. **Outbound reminder calls** — Virtual Reception enterprise; park until inbound is paying.

---

## What we will NOT copy

- Human-minute packs (Ruby/AnswerConnect) — wrong cost structure for $199 flat.
- Per-call add-on menus (Smith) — kills trust on “one price.”
- Free forever AI tier — see `receptionist/docs/MARKETING-PLAN.md`; free never tests $199.
- Fuller production number anywhere in demos or outreach.

---

## Build handoff (Helen / Claude Code — TM478)

- Keep demo line **+1 (650) 476-2005** only.
- Prove P0 items on general demos + 5 verticals; update `/projects/ai-assistance/` when live.
- Unique professional page designs = **TM483** (separate). This research page is content, not a redesign.