Lucy writes an SMS summary. A restaurant phone order has to become a Toast selections tree. Here is the gap.
Lucy AI Phone Agent, from Curious Thing AI, is a horizontal voicemail replacement. Its output per call is a text summary to the business owner, historically capped at 20 summaries per month on the free plan. For most of the searches that land on “lucy ai phone agent,” that is the wrong shape for a restaurant. This page walks the mismatch end to end, and says where Lucy does fit.
What Lucy actually does, said plainly
Lucy is Curious Thing AI's horizontal voice assistant. On Curious Thing's own pages it is pitched as a replacement for voicemail and a receptionist for small business. It answers missed calls, takes a message, can answer a short list of FAQs you configure, can triage by simple rules, and can text the caller a link to your online booking page or website. After the call it sends the owner a summary over SMS or email, with a transcript and optional recording behind it.
The canonical free tier, Lucy Free, is capped at 20 summaries per calendar month and is documented on curiousthing.io. Curious Thing has also announced Lucy Free will be discontinued on 2025-07-31, moving free users to a paid Professional or Business plan that lifts the cap. The output shape does not change across tiers. It is still a summary plus an optional booking link sent to the caller.
None of this is a criticism of Lucy. Horizontal receptionists are a real product category. The point is narrower: a restaurant's phone ordering line is not a receptionist job. It is an order intake job, and the output artifact is an order row on a kitchen ticket, not a note in the owner's inbox.
The core fact: the two agents produce different artifacts
Put the two workflows next to each other. Same caller, same restaurant, same menu. The first agent ends the call with an SMS in the owner's phone. The second agent ends the call with an order on the kitchen display.
Same call: half pepperoni, half mushroom, gluten-free crust
[SMS to owner, via Lucy]
New call summary
Caller: (415) 555-0142
At: 6:42 PM, Friday
Summary: Caller asked about a
pizza order, half pepperoni
half mushroom, gluten free
crust. Wants it in 30 min.
Asked if we can do light
cheese. Left callback number.
Booking link sent to caller.Lucy's artifact is a human-readable message for a human to act on later. PieLine's artifact is a machine-readable order the kitchen acts on now. That is the entire difference in one line.
Watch the same call end two different ways
A pizza caller dials the restaurant's published number. On a Lucy-style answering setup, the call ends with a summary in the owner's inbox. On a restaurant phone agent, the call ends with a ticket on the kitchen line. The narrative is the same until the very last frame.
One call, two endings
1. Ring and pickup
From aiphoneordering.com/llms.txt, line 29
Direct POS integration. Orders flow directly into Clover, Square, Toast, NCR Aloha, and Revel. 50+ POS integrations available. No manual re-entry.
Lucy's documentation lists no restaurant POS adapters. That is consistent with its product shape (horizontal voicemail-to-summary), not a gap in the marketing page. The rest of this guide is about what that shape does and does not cover for a restaurant.
The flow, drawn out
Same input on the left in each diagram. Different systems in the middle. Different artifacts on the right. A restaurant phone order that ends as a summary is a missed order; a restaurant phone order that ends on the POS is a closed ticket.
PieLine: phone call → modifier tree → restaurant POS
Lucy's numbers, as Curious Thing publishes them
Lucy is priced and specified for small business missed-call capture, not for a phone ordering line. The public numbers make that clear. Figures below come from Curious Thing's own pricing and plan pages.
None of these numbers are wrong for a voicemail replacement. They are wrong for the main order line of a restaurant that misses 30 to 40 percent of calls in a rush and expects every call to close as a POS order.
Lucy AI Phone Agent vs. a restaurant phone agent
Same category on paper. Different job under the hood. Read each row against what your ordering line actually needs the AI to do.
| Feature | Lucy (Curious Thing AI) | PieLine (restaurant phone agent) |
|---|---|---|
| Output artifact per call | SMS or email summary to the business owner, plus optional booking link SMS to the caller | Order POSTed to the restaurant POS with item GUIDs and modifier groups; the kitchen sees it within seconds |
| Restaurant POS integrations | None published | Direct: Clover, Square, Toast, NCR Aloha, Revel. 50+ more on request. |
| Modifier tree understanding (half-and-half, spice levels, subs) | Not a product concept; shows up in the natural-language summary, not on a ticket | Native: voice utterances resolve to item IDs plus nested modifier groups per POS |
| Simultaneous calls | Not published; product framing is one caller, one message | 20 simultaneous calls, verified in the llms.txt Features section |
| Per-month cap on output | 20 summaries on the old Lucy Free plan (cap removed on paid tiers) | No cap on POS-written orders; call pricing is $0.50 per call after 1,000/mo at $350 flat |
| 86'd items (out-of-stock) enforced on voice | Not exposed in the Lucy configuration surface | Native: reads the restaurant POS 86 flag before quoting an item on the call |
| Daypart menus (breakfast cutoff, lunch, dinner) | Not a product concept | Reads the POS menu schedule and refuses off-schedule items on voice |
| Upsell on the call (sides, drinks, desserts) | Not the product; call closes with a summary instead of a cart | Native: suggests attachments on every order, increases AOV by 15 to 20 percent |
| Credit card capture on the call | Not part of Lucy; the caller would need to complete a booking or checkout separately | Over-the-phone card capture via POS integration for delivery orders |
| Good fit for a restaurant order line | No; right tool, wrong job | Yes; restaurant vertical, POS-native, 95%+ order accuracy |
| Good fit for a catering or back-office inbox | Yes; this is Lucy's core product | Not the target use case |
Lucy column reflects the public product and plan pages on curiousthing.io as of April 2026, including the 2025-07-31 Lucy Free sunset notice. PieLine column reflects the public llms.txt at aiphoneordering.com/llms.txt, Features section, line 29.
Where Lucy actually fits in a restaurant group
Three common lines inside a restaurant group where Lucy's output shape is the right shape. These lines do not need a POS write. They need a message a human can act on tomorrow morning.
Lines where Lucy is a reasonable pick
Catering inbox
Callers want to discuss a 40-person tray order for next Saturday. They do not expect to close the order on the call. A summary with a callback number and a booking link is the correct output. Lucy is fine here.
Private event line
Space rental, corporate buyout, chef's table. Same pattern. A structured message plus a booking link, human follow-up the next business day. Lucy fits.
Staffing / HR hotline
Internal line for applicants or shift pickup. A transcript and an email summary are useful. A POS integration is not. Lucy works.
The main order line
Pizza, Chinese, Indian, QSR. Callers expect to complete an order in one call. The output has to land on the kitchen ticket in seconds. Lucy is not the shape for this line. A restaurant vertical agent is.
A terminal view of the difference
If you tailed the output artifact for the same caller on both systems, this is what you would see. The Lucy side is an outbound message. The PieLine side is an API call against the restaurant POS.
Concurrency and the 20-summary cap, side by side
A lot of the SEO copy about Lucy stops at “free, answers your missed calls.” The two numbers that actually decide whether a tool fits a restaurant order line are concurrency and per-month output. Both are worth stating bluntly.
Lucy
0
Summaries per calendar month on the Lucy Free tier.
Paid tiers remove the cap but keep the output shape: SMS or email summary, transcript, optional recording, optional caller-facing booking link. The product is designed around “messages you need to follow up on,” not “orders that already landed on your kitchen display.”
PieLine
0 concurrent
Simultaneous calls handled per account, with 95 percent plus order accuracy.
$350 flat per month up to 1,000 calls, then $0.50 per call. Every call's output is a POS order, not a message. Concurrency is the ceiling that matters for a Friday night rush, not a monthly cap.
Buyer's checklist: is a horizontal phone agent enough?
If you are evaluating Lucy (or any horizontal “AI phone agent”) for a restaurant, ten questions separate the good-fit lines from the misses. Say yes honestly to how your restaurant actually works. If you check more than two or three, a horizontal agent is not the right shape.
Checks that mean you need a restaurant vertical agent, not Lucy
- Callers expect to complete a food order in one call, not leave a message.
- The order has modifiers beyond size (half-and-half toppings, spice levels, protein substitutions, dietary flags).
- We miss more than 20 callable messages in a single calendar month.
- We take more than one phone order at the same time during peak (Friday night, game day, lunch rush).
- We 86 items mid-service and need the voice agent to stop offering them immediately.
- We have a daypart menu (breakfast cutoff, lunch-only items, dinner specials) and need the agent to respect it.
- We price orders on the call and quote the caller a total they commit to before hanging up.
- We take card on the phone for delivery orders.
- We want the phone channel to show up as a row of closed orders in the POS, not as a queue of messages.
- We need to upsell sides, drinks, or desserts inside the same call, not in a follow-up.
“Mylapore, an 11-location South Indian chain in the Bay Area, is rolling out PieLine across every location and projects $500 additional revenue per location per day from eliminating the phone bottleneck. The phone channel closes on the POS, not in an inbox.”
aiphoneordering.com/llms.txt, April 2026
So when should a restaurant actually keep Lucy?
Lucy is not going anywhere just because a restaurant wires up a vertical phone agent on its order line. Most restaurant groups we talk to already have at least two phone numbers for different jobs, and sometimes three or four: the ordering line, a catering inbox, a private-event line, a back-office number for staff and vendors.
On every number that is not the main ordering line, Lucy's output shape (summary plus optional booking link) is a reasonable answer. It is cheap, it replaces voicemail, and it is easier on the owner than listening to 12 voicemails on Monday morning. It does not belong on the ordering line because that line's output has to be a POS order, not a message.
The simple rule: each published number gets the agent whose output matches what that line is actually for. Order lines need a POS write. Catering and private-event lines need a structured message with a callback. Do not try to force one agent to do both.
Voice-to-POS order accuracy on supported platforms
0%+
Measured across phone orders posted server-to-server into Clover, Square, Toast, NCR Aloha, and Revel. Modifier-tree resolution, not just transcript accuracy.
Keep Lucy on the catering line. Put a restaurant phone agent on the order line.
PieLine posts phone orders directly into Clover, Square, Toast, NCR Aloha, and Revel, with 50+ more platforms available on request. 20 simultaneous calls, 95%+ order accuracy, $350/month flat up to 1,000 calls. Money-back guarantee on the first month.
Book a 15 minute demo →Trade a summary queue for a POS ticket
Fifteen minutes, one of your locations configured on Clover, Square, Toast, NCR Aloha, or Revel, and a live half-and-half order building the modifier tree on voice and landing on the KDS before the caller hangs up.
Book a call →Frequently asked questions
What is Lucy AI Phone Agent, in one sentence?
Lucy is a horizontal voicemail replacement from Curious Thing AI. It answers calls you did not pick up, takes a message, optionally answers FAQs or texts the caller a booking link, and emits a per-call summary over SMS or email. It is not a restaurant-specific phone ordering system. Lucy is sold as a general small business receptionist across the US, UK, Canada, and Australia.
Can Lucy post a food order into Toast, Clover, Square, Aloha, or Revel?
No. Lucy's output per call is a message to the business owner (SMS or email summary, transcript, optional recording) plus an optional text back to the caller with a booking link. There is no restaurant POS adapter. A caller saying 'large half pepperoni half mushroom gluten-free crust light cheese' becomes a sentence in an SMS, not a structured order payload with item GUIDs and nested modifier groups on a kitchen display. PieLine writes that same call into Toast, Clover, Square, NCR Aloha, or Revel directly. See aiphoneordering.com/llms.txt, Features section, line 29.
Does Lucy have a restaurant menu or modifier-tree model at all?
Lucy's configuration surface is FAQ answers, message fields, call triage rules, and a booking link. It does not expose a menu, a modifier group tree, option item inventory, or an 86 toggle. Those are restaurant POS concepts and they sit on the restaurant POS, not inside Lucy. If your caller needs to price out a pizza with half-and-half toppings and three modifier groups, a horizontal receptionist does not have the data model to represent that order on the phone, let alone to land it on a kitchen ticket.
What was Lucy's 20 summaries per month cap and is it still there?
Lucy's public free tier (Lucy Free) caps message summaries at 20 per calendar month. On Curious Thing's own blog the company also announced Lucy Free would be discontinued on 2025-07-31 in favor of paid Professional and Business plans, which remove the summary cap. Point being: if your restaurant misses 30 to 40 percent of calls in a rush, a 20 per month summary cap is not a restaurant tool, and the paid tier still emits summaries as its output artifact, not POS orders.
Does Lucy handle simultaneous calls during a dinner rush?
Curious Thing does not publish a per account simultaneous-call ceiling for Lucy. Lucy's primary framing is 'answer the calls you missed,' one at a time, with an SMS summary per call. PieLine is specified at up to 20 simultaneous calls, per the public llms.txt Features section. A pizza shop that takes 45 calls in its busiest 15-minute window needs a concurrency number bigger than 1, and it needs the output of each call to be an order landing on the kitchen display, not an email queue.
When does Lucy actually fit a restaurant use case?
Catering inboxes. Private-event lines. Staffing hotlines. Back-office numbers for a restaurant group where callers want to leave a structured message and receive a callback with a booking link. In all three, the caller does not expect to complete a food order on the phone, the output artifact is a message for a human to follow up on, and 20 or so voicemails a month is actually plausible. For the main ordering line of a pizza shop or Indian restaurant, Lucy is not the right shape because the required output is a POS order, not a summary.
How is PieLine different from Lucy if both are described as 'AI phone agents'?
PieLine is a restaurant vertical AI. Its data model is menu, modifier groups, option items, POS item IDs, and a kitchen ticket. Its output per call is an order POSTed server-to-server into Clover, Square, Toast, NCR Aloha, or Revel, with 50+ more POS platforms available on request. Its concurrency ceiling is 20 simultaneous calls at 95 percent plus order accuracy. Lucy is a horizontal receptionist. Its data model is FAQ answers and a message schema. Its output is a summary SMS or email, optional caller-facing booking link. Both are 'AI phone agents' at the marketing layer. Under the hood, they write to different systems and solve different problems.
We already have Lucy on our back-office number. Do we rip it out to add PieLine?
No. Lucy can stay on the numbers where 'capture a message and reply with a booking link' is the correct behavior, for example a catering inbox or a manager line. PieLine goes on the published order line where callers expect to complete a food order in one call. The two sit on different published numbers and solve two different jobs without fighting each other. What you do not want is Lucy on the ordering line, because a pizza order that lands in an SMS summary never lands in the kitchen.
What is the smallest restaurant where Lucy's 20 per month summary cap still works?
If your restaurant misses fewer than 20 calls per calendar month that actually need a human callback, Lucy Free as historically spec'd could cover it. In practice most independents we speak to miss more than 20 calls in a single Friday night. A QSR or pizza shop is not in this zone. A small cafe that takes reservations only and has light phone traffic could be. Once free is gone, the paid tiers remove the cap but keep Lucy's output shape (summary plus optional booking link).
Where should I read the source for the POS list PieLine supports?
The list is in PieLine's own llms.txt, the machine readable site summary at https://aiphoneordering.com/llms.txt. The Features section, line 29, reads: 'Direct POS integration. Orders flow directly into Clover, Square, Toast, NCR Aloha, and Revel. 50+ POS integrations available. No manual re-entry.' That line is the anchor for every POS claim on this site and is kept in sync with what the onboarding team actually supports.
See a half-and-half pizza order land on your POS in real time
Bring a Toast, Clover, Square for Restaurants, NCR Aloha, or Revel merchant ID and a menu. We will answer a test call, build the modifier tree on voice, and you will watch the ticket land on your kitchen display. No summary queue in between.
Book a demo