One Test Logged 150+ Contacts in 36 Hours. Only 1 in 10 Got a Rep.

Convoso said Tort Intake Professionals put its voice agent through a 36-hour trial and logged more than 150 contacts (Unite.AI). Convoso's no-code Voice AI Agent is now broadly available. The company announced that on November 3, 2026 (Unite.AI). It qualifies leads and moves sales-ready prospects to human reps (Unite.AI). In that trial, roughly 10–12% passed validation and went to human specialists in a warm transfer (Unite.AI).

What Convoso just put on the market

Convoso announced general availability for its no-code Voice AI Agent on November 3, 2026 (Unite.AI). The company dates the general availability to November 2, 2026 (Unite.AI).

Put simply, the agent handles the first conversation with a lead and decides how ready that person is before a human rep takes over (Unite.AI). For inbound calls, it can cover common questions, set up appointments and send the caller to the right team (Unite.AI).

Because it runs inside Convoso itself, it uses the same campaign logic, reports and compliance settings as your human outbound team (Unite.AI).

Convoso puts response time at about 600 milliseconds and says the agent learned from 20 million sales conversations (Unite.AI). It is built to handle callers cutting in, which matters in a real sales back-and-forth (Unite.AI). Buyers can select among multiple AI models and text-to-speech tools. They can pick from more than 700 voices, tune emotional tone and pace, and add voice cloning if they want (Unite.AI).

The agent is designed to hear short acknowledgements and keep going, work through objections, remember earlier parts of the call, and move between languages in a multilingual conversation (Unite.AI). My read: that matters most if many of your callers speak Spanish first.

A few features are worth a look. Auto Workflow can take old call recordings and help build qualification steps, so you do not have to hand-script every branch (Unite.AI). A browser test lets you talk to the agent before it takes live calls (Unite.AI). If the main model fails, a fallback chain moves the call to a backup model (Unite.AI).

One detail decides who can use it today. The agent connects only to Convoso's Dialer Platform through SIP/RTP, so for now it works with Convoso's own dialer stack (Unite.AI). My read: that limits who can try it right now.

The 36-hour test, in full

The first number Convoso shared comes from Tort Intake Professionals. Convoso said that group ran a 36-hour trial at the start and logged more than 150 contacts in that window (Unite.AI). Roughly 10–12% passed validation and went to human specialists in a warm transfer (Unite.AI). The team planned to take the testing further into advanced lead screening (Unite.AI).

Skip the easy read. Do the math instead. About one in ten. About one in ten ended in a warm transfer to a specialist, based on that 10–12% figure.

That is not automatically bad news. The agent's job is the repetitive first pass, and a warm transfer is the prize. But if your first reaction to that contact count was dollar signs, this is the correction. Volume is the easy part. The handoff is where the money is, and in this test the handoff was the exception, not the rule.

Nicholas Ward, who directs business operations at Tort Intake Professionals, said his group fields tens of thousands of outbound calls every day (Unite.AI). Ward described the agent as making the first connection, then pulling a specialist in at the point where that person matters most (Unite.AI).

The inbound story is closer to a dealership floor. Ace Building Materials is a family-owned supplier that has been in business for 75 years (Unite.AI). Five salespeople at the front counter were trying to serve people in the store and answer the phone at the same time, so other callers waited in line (Unite.AI). Its agent, called George, now handles common questions about hours and products and sends a caller to dispatch or another employee when the request is too much for it (Unite.AI).

George can tell when a caller is speaking Spanish and switches out of English on its own (Unite.AI). Josh Zakarya, the company president, said setup took about two weeks with help from Convoso (Unite.AI). He adds product and stock data, plus recordings of earlier customer calls, to improve the answers (Unite.AI). At Ace, George covers overflow calls so the sales staff can stay with the customers standing in front of them (Unite.AI).

Why the Ace example is the one to copy

The Ace example is the whole story. Five counter people, two jobs at once, a queue. Swap in a parts counter, a service desk, or a golf cart showroom on a Saturday morning and it is the same picture.

The early customer stories put the agent in two spots in the call flow: heavy volume first-pass qualification at Tort Intake Professionals, and overflow service at Ace (Unite.AI). In each case, the handoff to a person is the key part — specialists get prospects that passed validation, and employees get callers whose needs go beyond a routine answer (Unite.AI).

The write-up says the launch matters because those handoffs, reports and controls all sit in one Convoso operation (Unite.AI). My read: that is the difference between a demo and a workflow you can run.

What this means for you if you sell powersports or golf carts

Here is my read, and it is opinion, not reporting.

The interesting number is not the 150. It is the handoff rate. An agent that talks to everybody and hands over one in ten is not replacing your sales team. It is triaging for them. Triage is worth real money on a Saturday.

Picture a powersports dealership running a spring ATV promotion. Your people are on the floor with customers, the phone keeps ringing, and nobody picks up. My read: many of those calls are the routine kind the agent is built for. In the Ace setup, the agent takes that overflow, and your reps stay with the customers in front of them (Unite.AI). Your floor does not empty out to answer the phone.

Picture a golf cart dealer in a market where a lot of buyers speak Spanish first. The agent knows Spanish and moves over from English by itself (Unite.AI). My read: a caller who starts in English and switches mid-sentence is more likely to stay on the line, and on a weekend that could be the difference between a demo booked and a lead that calls your competitor next.

Now the third scenario, and the one that stings. Picture a multi-location group that hears about 150 contacts in 36 hours and buys the volume promise. If the qualification logic is sloppy, you get a pile of contacts, a thin slice of real buyers, and a sales floor that stops trusting the tool by week two. The sources give no promise about your handoff rate. You test it small or you own the mess.

How I would use this week

1. Find out if you can even run it. The agent is tied to Convoso's Dialer Platform over SIP/RTP, so for now it only plugs into the Convoso dialer stack (Unite.AI). If you are not on that dialer, you cannot deploy it right now. Treat the rest of these steps as a benchmark you can copy with whatever you have, and a checklist for the next time you shop phone systems.

1. Pull a week of your own recorded calls. This matters because Auto Workflow can use past recordings to help shape qualification steps (Unite.AI). Even a stack of your own recordings tells you what people actually ask in the first thirty seconds. Thirty minutes with a notepad, that is all it takes.

1. Write down your most-asked questions. Store hours and product questions. Those are the routine kind an inbound agent is meant to field (Unite.AI). Anything past that gets routed.

1. Decide who receives the handoff, by name. The whole value sits in the warm transfer. If the transfer lands in a general voicemail box at 11 a.m. on a Saturday, you paid for nothing.

1. Set the after-hours rule before the busy-season rule. My read: after-hours is the cheapest win on the list.

1. Test it in the browser before it touches a live caller. A browser test lets you speak with the agent before it takes live calls (Unite.AI). Use it. Try to break it with the questions your worst customer asks.

1. Run your own 36-hour test and count handoffs, not contacts. Convoso said the Tort Intake trial ran for 36 hours (Unite.AI). Count how many calls ended with a real appointment or a real transfer. That is your number. Contacts are vanity.

If you want a starting script to hand the agent, here is roughly what I would type in for each side of your business. For a powersports dealer, I would have it ask how soon the caller wants to buy and how they plan to pay. For a golf cart dealer, I would have it ask whether the cart is for a home, a course or a rental fleet, and whether it needs to be street legal. For an overflow greeting on either one, I would have it ask whether the caller is checking stock, booking service or asking about a price, then route from there.

Mistakes I see owners make with this stuff

• Buying volume and forgetting the handoff. The Tort Intake Professionals test ended with about 10–12% passing validation and going warm to human specialists (Unite.AI). If your own handoff rate is worse than that, the tool is not your problem. Your qualifying questions are.
• Not checking your dialer first. For now the agent only works with the Convoso dialer stack (Unite.AI), so if you are not on it, this is not a this-week project.
• Pointing AI at your best leads first. Let it take the overflow and the after-hours calls first.
• Never listening to the recordings. At Ace, Zakarya adds product and stock details and past call recordings to improve responses (Unite.AI). Feed it nothing, get nothing.