Voicebox, a startup whose existing customers include airport terminals, has launched a "catalog" that lets someone record a spoken review of any place from anywhere, rather than only from a tag on the counter. Until now the product did one thing: a customer scans a QR code in a store or taps their phone against an NFC tag, talks for as long as they want, and the service transcribes the recording, pushes it to the company's dashboard and attaches a sentiment analysis. If the company wants to follow up, it can do so by email. The catalog removes the tag from the loop, and in future versions users will be able to find other people's public opinions about a specific company. That is a different business from the one Voicebox has been in.
The case for voice feedback rests on an asymmetry in effort. Most people never write to customer support unless something has gone seriously wrong and they are trying to get money back. Twenty seconds of talking into a phone — it went well, it went badly — costs almost nothing by comparison, and less than waiting an unknown amount of time for an agent to pick up.
Voicebox chief executive Karan Gupta says voice technology passed an inflection point over the last two years: accurate enough, and fast enough. He has been here before. His previous startup also built on voice, an app called Alice for secure speech transcription.
The wider evidence is on his side. Some of the most visible software of this year is built around recording the voice. Wispr turns speech into accurate dictation. Granola listens continuously to what is happening around it. Dictation has become good enough that drafting in a notes app and then reading it aloud to a laptop is a normal way to write. Some forecasters have declared this the end of the keyboard era, with office workers soon murmuring into desk microphones instead of typing. That prediction asks too much of the technology. The narrower claim — that voice changes how customers leave feedback — asks much less.
In airport terminals, the existing deployment, travellers can report how dirty the toilets were or rate how legible the signs to the exits are. These are complaints nobody would ever type. That is the point, and it is a real one.
Gupta wants spoken feedback to go beyond dealing with companies. Earlier this year Lena Dunham held an event in San Francisco for her book Famesick and put a Voicebox QR code on the screen. A lot of attendees left messages, including emotional ones about relating to her experience of illness. Gupta treats this as evidence that people want easier ways to share opinions about things they care about.
That episode demonstrates less than it appears to. A room full of people who came to see an author, pointed at a code on a screen, will leave messages. That is a captive audience, not demand discovered in the wild. The catalog needs the wild.
And the catalog is where the ambition sits. It moves Voicebox from a tool for collecting customer complaints toward something closer to Google Maps with the reviews spoken aloud. This fits the broader direction of software right now — dictation feels natural enough that thumb-typing on a phone starts to look clumsy and dated. It also moves the company into the hardest category of consumer product there is. A review network is worth nothing until it has density, and density requires people to open it when nothing is wrong, which is exactly the behaviour the current product does not need and has never had to prove.
Notably absent from the launch is anything about the part that breaks review networks rather than builds them. Who moderates a public voice review. Whether a business can dispute one. What stops the catalog filling with the same spam that fills text review sites, now with the added cost of transcription and storage on every submission. Voicebox's existing business is a dashboard sold to an airport; the catalog is a public square, and public squares are governed or they rot.
There is also a friction the product does not remove. It takes the keyboard out of the way, and leaves the audience in. Whispering your opinion of a haircut into a phone is easy right up to the moment someone is sitting next to you, and the natural instinct is to wait until they have gone. The obstacle to leaving a review was never only the typing.