Hi Auto's $15M and the Quiet Consolidation of Voice AI's 'Second Tier'

I was standing in the parking lot of a Bojangles outside Charlotte this morning, holding a cup of coffee that had gone lukewarm somewhere between the third and fourth car in the drive-thru line, when the Hi Auto press release hit my inbox. The order ahead of me — a four-combo with two substitutions, a “no ice, extra ice on the side” caveat I’m still trying to parse, and a kids’ meal upsell that the bot tried twice — had just been handed off to a human at the window. I watched the manager mouth “sorry about that” through the glass while the AI did its little chirpy goodbye in the speaker post above my head.

So: same morning, same Bojangles, two data points. One: the voice AI took the order. Two: Hi Auto, the company that almost certainly built it, just raised $15M.

That’s the kind of coincidence that makes me write a column instead of finish my biscuit.

Vibe Check verdict (above the fold)

• Build quality: 7/10 — production-grade where it matters, still rough on edge cases.
• Operator impact: 7.5/10 — measurable lift on labor and accuracy at the anchor accounts.
• Strategic positioning: 9/10 — this is where the round actually matters.
• Hype-to-substance ratio: Better than most.
• Watch-list status: Top of the second tier. The most interesting voice-AI vendor in the QSR stack this quarter that you’re not already arguing about on LinkedIn.

The TL;DR for operators: if you are scoping a drive-thru voice pilot in Q2 and your shortlist is SoundHound, Presto, and “the rest,” Hi Auto belongs in the conversation as a serious option — not as a long-shot.

Now let me show my work.

What Hi Auto actually announced

The numbers, per the PRNewswire release:

• $15M Series A, closed and announced today.
• Round led by Delek Motors, with participation from the Zisapel Family, Vasuki Tech Fund, and an undisclosed publicly-traded QSR strategic.
• Total raised across the company’s history: $23M.
• CEO Roy Baharav anchors the release with a quote about labor turnover — specifically, that voice AI is being pulled into drive-thrus less by operators chasing a futuristic experience and more by operators who simply cannot keep an order-taker in the headset for more than six months.

That last point is the part of the release I want to underline. The press copy frames Hi Auto as a labor-shortage answer first and a guest-experience play second. That ordering matters. It tells you what the company is selling and to whom — which, judging by the customer list, is operators who are tired of the headline AI demos and want a thing that takes orders at 11pm on a Tuesday without quitting.

The client roster as disclosed:

• Bojangles
• Checkers & Rally’s
• Lee’s Famous Recipe
• Burger King New Zealand
• Popeyes UK

Two domestic regional chains with real footprints. One domestic combo brand with two banners and a heavy late-night mix. Two international flagship licensees of brands you’ve heard of, in markets where the parent brand isn’t necessarily steering the tech stack day-to-day.

That is a remarkably coherent customer profile, and we’ll come back to it.

Who’s actually paying Hi Auto

Here’s the thing about the voice-AI category in 2025: the customer logos do most of the talking, and almost nobody is putting up the logos you’d actually need to see to make a real comparison.

You can read Restaurant Business’s recent piece on the voice-AI market getting crowded and come away with the impression that there are eight or ten vendors and they’re all roughly interchangeable, all racing to win the same two or three enterprise pilots. That’s directionally true and tactically useless.

What I look at instead is the named-anchor problem. A voice-AI vendor has three kinds of customers in their deck:

1. Named anchors: brands they can list on their homepage, in press, on a conference panel. Real deployments, real contracts, real lane-count.
2. Pilots-in-progress: brands they can refer to obliquely (“a top-10 burger chain”) but cannot name yet. Some of these will convert. Many will not.
3. Aspirations: brands they have had one or two conversations with at a trade show. These show up in pitch decks and never in revenue.

The whole game right now is what fraction of a vendor’s pipeline is category one. And by that measure, Hi Auto’s five named anchors — three of them domestic, including a multi-banner combo operator — is not a tiny list. It’s not Yum-scale. It’s not McDonald’s. But it’s enough chains, enough lanes, and enough geographic and menu diversity that the product has had to handle real-world variance and not just demo-bench variance.

Build quality

• Trained on enough menu and accent variance to handle real Southeast US drive-thru traffic at Bojangles and Lee’s volumes.
• Combo and modifier handling is the published strength — exactly the part that makes or breaks late-night accuracy.
• The remaining weak spots are the ones every voice-AI vendor has: substitutions, “no [item] on the [other item],” and the long tail of regional menu hacks customers invent on the fly.

I’d put Hi Auto’s product, today, somewhere in the same band as the loudest names in the category for the 80% of orders that look like normal orders. Where it pulls ahead is the customer composition, not the demo reel.

What “second tier” actually means here

I want to be careful with this phrase because it sounds like a slight and it isn’t.

In the voice-AI conversation, “first tier” gets used to mean the vendors that are either publicly traded (SoundHound) or that have eaten the most oxygen in the trade press (Presto, with its public-company drama and its own running narrative around the Checkers contract on its product site). The “second tier” is everyone else who is actually shipping into real lanes — Hi Auto, ConverseNow, Kea, OpenCity (more reservations than drive-thru, but adjacent), and a few specialty players doing audio-IVR-style work that gets quietly white-labeled.

What I’m seeing in Q1 is that the second tier is where the contract activity is. The first tier is where the headlines are. Those are not the same thing. The press cycle around any given SoundHound earnings call or any given Presto board fight tells you almost nothing about who is winning the next 200 drive-thru lanes.

Hi Auto’s round, and specifically the shape of the round, is one of the cleaner pieces of evidence I’ve seen this year that the second tier is consolidating into “vendors with at least one named multi-unit anchor” and “everyone else.” Hi Auto is in the first bucket. Several of its nominal peers, by the named-anchor test, are not.

This matters for operators in a specific way: if you’re a 200-unit operator scoping a voice pilot, the question is not “which vendor has the best demo.” It’s “which vendor will still be answering my phone in 2027 when my menu changes and I need a new prompt set.” Vendors with three or four named anchors and a recent Series A are more likely to be that vendor than vendors with one anchor and a seed round that closed in 2023.

The undisclosed strategic and what it implies

This is the part of the press release I’ve been chewing on all morning.

A publicly-traded QSR strategic put money into this round and didn’t put their name on it. That’s a choice. There are roughly three reasons a strategic does that:

1. They’re piloting and don’t want to commit publicly until the pilot lands. Most common. Implies a defined evaluation window — usually 6 to 12 months — at the end of which the strategic either signs a multi-banner deal and lets Hi Auto announce, or quietly walks.
2. They’re hedging. They have another vendor in the lane today (likely one of the first-tier names) and they’re keeping a Plan B funded so they have leverage at renewal.
3. They’re protecting a competitive position. They don’t want their competitors to know they’ve made a voice-AI selection at all.

I don’t know which of those three is in play. What I do know is that the disclosed investors — Delek Motors (yes, the Israeli vehicle and energy group), the Zisapel family (a serious Israeli tech-investing pedigree, the Zisapels have backed roughly a generation of Israeli networking and AI companies), and Vasuki Tech Fund — do not, on their own, get you to $15M at this stage in this category. The strategic check is doing real work in this round, not just sitting on the cap table as a friendly observer.

If I had to guess, and this is reading tea leaves: the strategic is probably one of the larger franchisor-style publicly traded QSRs that has been quiet about voice in earnings calls. That narrows the field, but I’ll leave the speculation to people who have less to lose than I do.

Operator impact

• If you’re evaluating Hi Auto, ask explicitly whether you’d be in the same pilot cohort as the strategic. Cohort dynamics matter — the strategic’s lane count will get priority on roadmap.
• The press release’s labor framing is the right framing. Don’t let a salesperson re-pitch this as a guest-experience play. The ROI math that closes is the headcount math.
• The international anchors (Burger King NZ, Popeyes UK) are good evidence of multi-language handling, but don’t assume the US English models and the UK English models are the same thing. They’re not.

Where this leaves SoundHound and Presto

Quickly, because I want to come back to Hi Auto.

SoundHound has the public-market spotlight and the broader platform story — voice across automotive, hospitality, and QSR. That platform breadth is a real moat in some respects and a real distraction in others. If you’re a 50-unit operator, “broad platform” means “you are not the priority customer for the roadmap.” That’s not a knock; it’s a function of where their revenue concentration sits.

Presto’s narrative is messier. The product is real and shipping, the Checkers anchor is real, but the company-level story has spent more of Q1 in headlines about governance and capital structure than about deployments. If I’m an operator, I am not penalizing Presto for that — the lane-level execution is, by most accounts I trust, fine — but I am asking my account manager direct questions about runway, support continuity, and product roadmap ownership. Those are reasonable questions to ask of any vendor; they’re more pointed questions to ask of Presto right now.

Hi Auto’s position relative to both: smaller, less storied, more concentrated customer base, fresh capital, and — critically — an undisclosed strategic that has a structural reason to want Hi Auto to succeed. That last factor is the asymmetric one. SoundHound’s strategic backers are diffuse. Presto’s are public and contested. Hi Auto’s are quiet and aligned.

That’s a better shape of cap table for a vendor that needs to win the next 18 months on contracts, not on press.

What this means for the broader stack

I want to zoom out for one section, because the voice-AI conversation gets stuck inside its own category when the more interesting question is how it interacts with everything else in the restaurant tech stack.

A piece we publish later on the POS-side AI rollout walks through how the order-side AI (Toast Sous Chef and its lineage) sits relative to the rest of the operating stack. Voice AI on the drive-thru sits in roughly the same architectural position — it’s a front-of-house input layer that hands a structured order to a POS, kitchen-display, and inventory system that may or may not be ready for it.

The piece of this that operators most consistently underestimate: the integration tax. A voice-AI vendor can have a flawless model and still cost you margin if the handoff to your POS introduces a 2-3 second latency on every order, or if it can’t represent a modifier the way your kitchen reads it. Hi Auto’s anchors — Bojangles, Checkers, Lee’s — all run different POS environments. The fact that Hi Auto is shipping into all three is, by itself, evidence that the integration layer has matured past the demo stage.

And then there’s the margin math. A forthcoming framework piece we publish in our Mise column lays out the four margins of a restaurant — food cost, labor, rent, and a fourth we’ll get to in that piece — and voice AI is one of the very few tools that can move the labor margin without moving any of the other three in the wrong direction. That’s why this category gets the capital it does even when the press cycle is unkind to specific vendors. The macro thesis is right; the vendor selection is the hard part.

If the macro is right and the contract layer is bifurcating, then “which second-tier vendor has the right anchors and the right cap table” becomes the operating question of the year. Hi Auto’s round is the cleanest answer to that question I’ve seen since January.

Where Hi Auto could still fall over

I’m not in the business of writing fan letters, and Vibe Check exists specifically to put the doubts in the same room as the verdict. So:

Concentration risk on the anchors. Five named anchors is enough to make a credible case, but if any one of those — particularly Bojangles or Checkers — churns or pauses, the story changes fast. Voice-AI contracts are usually structured with explicit performance gates (order accuracy, transfer rate, escalation rate) and any vendor in this space lives quarter-to-quarter on those gates.

The strategic could go either way. Today’s quiet strategic backer is tomorrow’s quiet strategic walk-away. If the pilot doesn’t hit, the same logic that put the check in unmarked also pulls it out quietly. Operators evaluating Hi Auto should ask about cohort isolation — i.e., is your deployment dependent on the strategic’s pilot, or independent of it?

Roy Baharav’s labor-framing is correct but limiting. Selling “we save you on headcount” works in 2025 because the labor market is what it is. If the labor picture eases in 2026 — and there are reasons to think parts of it might — the framing has to evolve. Hi Auto needs to be ready to tell a guest-experience and throughput story when the labor story softens. I don’t yet know if they’re set up to do that.

International is harder than the deck suggests. Burger King NZ and Popeyes UK are great logos. They are not a multi-country product strategy. A vendor that wants to be a multi-region default needs language coverage, regulatory coverage (recording consent law varies wildly by jurisdiction), and a support footprint that none of the named second-tier vendors yet have.

Watch-list (3-4 bullets each)

For operators:

• If you’re scoping Q2/Q3 pilots, add Hi Auto to the shortlist. Don’t sign yet; do shortlist.
• Ask explicitly about cohort dynamics with the undisclosed strategic. You want to know if you’re getting priority or being deprioritized to the strategic’s roadmap.
• Get the labor-savings number in writing and tied to a measurement window. The press release leads with labor; your contract should too.

For investors:

• Hi Auto’s $23M total raise gets them to the next inflection if conversion at the named anchors holds. It does not get them to a platform exit on its own.
• The strategic’s identity will become known eventually, either through deployment announcements or — more likely — through a follow-on round where they un-blind.
• Watch the Q3 trade-press cycle for the first named expansion. If Bojangles goes from “x lanes” to “all lanes” between now and September, the round is validated.

For the rest of the category:

• SoundHound and Presto are not Hi Auto’s direct competitors at this moment. ConverseNow and Kea are.
• The bifurcation is happening on the contract layer, not the product layer. Vendors who can’t show three named multi-unit anchors by year-end will struggle to raise.
• Watch which strategics get loud and which stay quiet. The quiet ones are the leading indicator.

Vibe Check call

Hi Auto’s $15M is not a “company-defining” round in the sense that the press release would like you to believe. It’s not a Series C, it’s not a $100M number, it’s not a unicorn moment. What it is, and what makes it worth a column, is a structurally clean round: real customers, a coherent customer profile, a strategic check that signals contract-layer momentum, and a CEO framing that matches the actual buying motive in the market.

The voice-AI category in 2025 is going to be decided by two things: which vendors have named multi-unit anchors by the end of the year, and which vendors have strategic capital with skin in the deployment game. Hi Auto, today, is in the small group that has both.

So the Vibe Check call is: not the loudest, not the biggest, but the second-tier voice-AI vendor most worth watching this quarter. If you’re an operator with a Q2 or Q3 pilot RFP open, put Hi Auto on the list. If you’re an investor with category exposure, this round is a signal worth re-reading the rest of your portfolio against. If you’re a competitor, the bar to look credible at year-end just went up — and you got 12 days’ notice that it did.

I’m going back to my lukewarm coffee and watching the order screen at the speaker post above me cycle through its menu loop. The voice in the speaker is reading the day’s specials in a tone that is, I have to say, more cheerful than the manager inside the window. I’ll take it.

Vibe Check verdict: Hi Auto is the most interesting “second-tier” voice-AI vendor in QSR this quarter. Shortlist material for operators; signal material for investors. Not yet platform-defining; meaningfully more credible than it was yesterday.

— Sofia runs Vibe Check. Tips: tips@tabletransfers.com.