The Reycor Review — Week of November 5, 2026
The Reycor Review
Week of November 5, 2026
The owners in this issue had the same kind of leak you can point to on a calendar: calls after the desk closed, reviews that did not follow a new shop, invoices that never got typed. They used AI to catch the work. They also left a person in charge of anything a customer would rely on. Where the numbers come from a vendor, or from an owner talking to that vendor, we say so.
The salon that was losing after-hours shoppers
Cindy Reynolds runs Mermaid Hair Extensions, a two-location salon near Seattle. Extensions buyers often shopped after the desk had closed. Even when the salon was open, the front desk could not catch every call, and the questions were the same ones, over and over: how long does it take, which method, how long does it last, what does it cost, and will it damage my hair.
Reynolds put those answers into TrueLark, an AI receptionist tied to the salon’s Phorest booking software. When a call is missed — closed, or just busy — TrueLark texts the caller: “This is Mermaid Hair Extensions, and we are sorry we were unavailable to take your call, but how can we help you?” It can book, cancel, or reschedule in the system the desk already uses. Staff still try to pick up during open hours. The team sometimes prefers to handle reschedules by text.
Reynolds told Salon Today she “was surprised to learn how many people were scouting our salon after hours.” The figures she and the article report for 2023: 1,188 new clients first booked through TrueLark; an average of 28 online bookings a week, 17 of them after hours; and about $26,000 more revenue a month. That is owner- and article-reported, not an independent audit. The August 9, 2024 piece also quotes TrueLark’s CEO at length, so read it as trade press with a vendor in the room.
The same article describes Weena Jerome-Alexandre at Renaissance Curls in Harlem, using Boulevard and TrueLark. The tool fields 82% of incoming calls; she can watch the answers and jump in. She reports about 30 new clients a month, roughly $2,400 in extra revenue.
If your phone is the front door after 6 p.m., the move worth copying is the one Reynolds made: write down the five questions you already answer, connect an after-hours text-back to the booking software you already use, keep a person on the line when you are open, and read what the tool said before you treat it as done.
The repair company that needed reviews in every new market
Valentin Pashnyak founded Appliance EMT, an appliance-repair company headquartered in Buford, in the Atlanta metro. As the company added locations, the reviews did not travel with them. A September 2024 snapshot on ReviewWheel’s customer page puts the Buford headquarters at 1,342 Google reviews and 4.7 stars. Pashnyak also says more than 900 reviews were stuck in another platform, BirdEye, and disappeared when that account was canceled. Missed calls and uneven follow-up on estimates made it harder to fill the calendar.
The stack they describe is simple: a QR code on day one, after the job, pointing customers to Google; AI replies within minutes (a personalized thank-you on five-star reviews, a more careful reply on critical ones); reactivation campaigns for past customers; and the same setup copied to each new shop. “We needed reviews to scale across new markets,” Pashnyak is quoted as saying. “The platform behind ReviewWheel handled every customer, every job, every text automatically…”
As stated on that page, from September 2024 to November 2026 the Buford shop went from 1,342 reviews to 4,717 and from 4.7 to 4.9 stars. The company reports 8,688 reviews across 14 locations, a 40% higher booking rate at headquarters, and growth from 5 locations to 14 with the same office staff. This is a vendor customer page. The company frames some of the revenue effect as “influenced,” and none of it is independently audited, so treat those figures as the vendor and owner’s account, not as a guarantee.
If you are adding a second truck or a second shop, the part worth copying is operational, not magical: ask for the review before you leave the driveway, reply quickly and in a human voice, and keep a person in the loop on anything below a five-star rating.
The books: let the tool draft, you still approve
A Perth bookkeeping firm, Digit, tested Xero’s auto-reconciliation (powered by JAX) on real client files from January through March 2026, for up to about ten weeks. Andrew Erkins published the notes on March 31, 2026. They turned it on account by account, reviewed matches in “View All Reconciled,” and were selective — leaving it on for credit cards, for example, and off for transfer-heavy accounts.
On a day-spa file, credited to Princess Espinosa, 69 of 116 transactions auto-reconciled in 30 days, about 60%, with no errors flagged. A subscriptions client hit 26 of 39, about 67%. On more complex AV and intercompany files, they saw false matches and had to remove and redo work weekly. Timing clashes with receipt-capture software (Dext) showed up too. Digit’s conclusion is the useful one: “The ‘set and forget’ pitch doesn’t hold up in practice. Not yet.” This is a firm writing up its own test, not an independent audit, but it is unusually plain about where the tool failed.
Two Xero partner bookkeepers described a similar rhythm on Xero’s blog on November 3, 2026. Katie Williams of The Vault Bookkeeping in Australia’s Hunter Valley started on simpler, secondary accounts, checked daily in the first week, then moved to fortnightly or monthly review, and kept manual bank rules for high-certainty items such as loans and known GST. She says she was “skeptical [at first] and I was also nervous,” and later: “AI is a tool we use to help us do the work we do.” She reports about six hours of productivity gain per employee, and says a monthly-to-quarterly package went from her worst return to a top revenue maker — qualitative, and published on Xero’s own site. Martine Hoosen of The Bookkeepers Academy says that after she corrected miscoded lines, the system adapted; in the last 30 days before that article, about 97% of statement lines auto-reconciled. Final say, both say, stays with the team.
Sandra, owner of PestAway NZ, a pest-control route business in Taranaki, New Zealand, had a different leak: every finished job had to be retyped into Xero, about five minutes each. She did the books herself, and some completed work never became an invoice. Fiord AI’s project write-up describes a custom agent that runs at 1 a.m., pulls the prior day from Fieldwork HQ, matches customers, line items, services, and GST, and creates Xero drafts for Sandra to approve. It also syncs contacts and was used to recover invoices that had been archived or mislabelled.
As of July 2026, after about a year in production (live since July 2025), the write-up reports 2,995 invoices synced, about $725,000 NZD of revenue moved through that pipe, 759 contacts synced, more than 250 bookkeeping hours eliminated, and about $6,000 NZD of unbilled work recovered, including an April 2026 recovery of 133 invoices worth about $4,400 NZD. Sandra is quoted: “It’s saved us hours every week and caught revenue that would’ve slipped through the cracks.” This is the implementer’s portfolio page — a named client, an owner quote, and vendor-measured numbers, with no independent audit.
The pattern across these stories is the part to steal: turn the tool on where the file is simple, correct its mistakes so it learns your codes, never let it file a payment or send an invoice you have not seen, and measure the hit rate on your accounts instead of trusting a demo.
What a careful owner does with any of this
AI only helps if your customers still believe you. If a text, a review reply, or a phone greeting is drafted by a tool, say so. Promise only what you can show. Put a named person behind anything a customer will rely on — a booking, a price, a reply to a complaint, a reconciliation, an invoice — and write down who is responsible when the tool gets it wrong.
The owners above did not hand the business to software. They used it to catch the missed call, ask for the review before leaving the job, and draft the invoice or the bank match. Then they checked the work.
Sources: Salon Today, August 9, 2024; ReviewWheel / Appliance EMT, timeline September 2024–November 2026; Digit, March 31, 2026; Xero Blog, November 3, 2026; Fiord AI / PestAway NZ, measured July 2025–July 2026.