If you run Revel, one question decides whether a voice agent is worth buying: does a finished phone order land in Revel as a correct ticket, priced the way the register would have priced it? Voice quality is the part every demo leads with and the least useful thing to judge. An agent that sounds excellent and drops a modifier costs you more than a call that rang out, because you make the food twice and the person on the other end was already your customer.
X1 Voice reaches Revel through Deliverect. Your published items, prices and modifier groups come across, the agent prices each call against them, and the completed order is written back as a normal order. Nobody re-types anything.
That routing is worth understanding rather than skipping, because it explains both what setup asks of you and where the system will occasionally get in your way.
What the Deliverect connection changes
A direct integration talks to the POS. A channel integration talks to middleware that talks to the POS. Three things follow.
Your Revel menu has to be published to Deliverect first. If you already push to delivery marketplaces, this exists and setup is short. If you don't, publishing the menu is the actual first task and it is the step most likely to add a day to the schedule.
Menu changes land on a sync, not on the next ring. A price change on that cadence is harmless. An item you 86 at 7pm on a Saturday is not, which is why you want a clear answer about how sold-out items are handled before you trust the agent to tell callers what is available. Voice AI and POS 86 sync failure modes covers what to ask.
Names come across as whatever they are in Revel. Menus that have been running for a few years accumulate item names built for a tap target rather than for speech: abbreviations, internal codes, sizes glued to the front. The agent inherits all of it and will say it aloud on a call. That is the most common reason a working agent sounds wrong on its first day, and it is a naming job rather than a software defect. Read your item list aloud once and you will find them in about ten minutes.
What it looks like on a real shift
It's 6:45 on a Friday. Two lines are ringing, the counter is four deep, and somebody wants to know whether the sauce has dairy in it.
The agent answers on the second ring, takes a pickup order with a substitution and a hold, charges the card, and the ticket prints at expo with the modifier attached. Nobody behind the counter looked up. A second caller asks about catering for thirty on Sunday, and the agent doesn't attempt it: it captures the name, number, headcount and date, transfers if a manager is on the floor, and leaves a transcript if not.
That second behavior is the one worth checking in a demo, and it is a configuration decision rather than a capability. Human handoff and failover covers how to write those rules so the agent hands off on the calls you actually want a person taking.
The gain shows up in calls you currently have no record of. A phone ringing out during a rush leaves no trace, which is exactly why most operators underestimate their own miss rate. Count a typical week's missed calls, multiply by your average ticket, and check the number before deciding the problem is too small to solve.
The failure mode nobody puts in a demo
Modifier structure, and it is worth being direct about it.
Menus grow by accretion. Somebody adds a half-and-half option to one item and not its sibling. Somebody builds three sizes as three separate items instead of a size group. A required modifier group carries a default that only makes sense with a person standing at the screen. None of this hurts you today, because the employee ringing it in knows what the buttons mean and works around the structure without thinking about it.
An agent has no such judgment. It follows the structure literally, so it will either refuse a combination that is obviously fine or fire a ticket that validates and is wrong in the kitchen.
The fix is structural rather than conversational, and for most menus it is a couple of hours of one-time work. Simplifying modifier trees covers the specific shapes that break.
So test for it. Ask any vendor to take a deliberately awkward order on your real menu, not a sample: "large, half one way half the other, no onions on the second half, sub the side, add a drink, apply the promo." Then walk to the printer and look at the ticket. Do not accept the agent repeating the order back correctly as evidence. The repeat-back and the injected ticket are two different systems, and only one of them feeds the line. That is why integration depth decides this purchase and voice quality does not.
The numbers to pull after a month
Four, and they take about twenty minutes.
Answered-call rate, which should approach 100 percent and is the reason you bought it. Order accuracy, measured by reading a sample of phone tickets against their transcripts rather than trusting a dashboard, using the order accuracy measurement method. Escalation reasons, sorted into "by design" and "by defect," where only the defect bucket is a to-do list. And phone revenue against the same weeks before you turned it on.
If answered calls went up and phone revenue did not, the agent is absorbing calls that staff were already picking up. That still buys you labor back, but it is a different case than incremental revenue, and you should know which one you are making rather than assuming the better one.
What setup actually costs in time
The connection is minutes. Reviewing what came across is the part that takes a person's attention.
Plan on someone sitting with the imported menu once, reading item names aloud, and fixing the ones that sound like database fields. Set hours, delivery zones if you deliver, and any phone-only rules such as a delivery minimum. Then place five or six test calls yourself and order the way your regulars talk rather than the way the menu is written. Most single locations are live well inside a day, and plans start at $250 a month, which you can size against your own call volume on the pricing page.
If you run Revel at some locations and something else at others, that is not a blocker. X1 Voice connects directly to Square, Clover and OrderCounter and reaches the rest of the supported POS list through Deliverect, so a mixed estate runs on one agent configuration.
The decision rule is small enough to run this week. Place five hard test calls on your own menu, walk to the printer each time, and count how many produced a ticket you would have been happy to see a new hire ring in. Five means the integration is doing its job. Three means the problem is almost certainly your modifier structure rather than the voice, and that is worth fixing before you sign anything.