The Company That Taught Radio to Talk Back

In February 2023, weeks into the global freakout over ChatGPT, a software company most people outside broadcasting had never heard of walked into radio's biggest anxiety and gave it a name. Futuri, based in Austin, announced RadioGPT: a radio host with no lungs, no coffee habit, and no contract to renegotiate. It could read a station's music log and talk about the songs, tease the next break, deliver the weather, and riff on whatever was happening in town that morning. Program directors across the country asked the same two questions at once. Is this real, and does it come for my job?

The honest answer to the first was: mostly, and improving fast. The answer to the second was more interesting, and it explains Futuri better than any product sheet. The company did not set out to replace radio people. It set out to delete the parts of their week nobody enjoys - the source-scanning, the script-drafting, the spec-ad grind, the prospecting - and to sell that time back to the station. The tagline on Futuri's site says it plainly: automate the grind, elevate the genius.

The setupA decade of listening first

Futuri did not start as an AI company, because in 2009 almost nobody did. Daniel Anstandig, who had been working in radio since he was a teenager, founded it in Cleveland with backing from the Zapis Capital Group. The first product was almost quaint by today's standards and radical by the standards of its day: Listener Driven Radio, a "vote for what plays" system that let an audience shape a live broadcast in real time. The premise was that a station could be a two-way medium instead of a one-way tower.

That instinct - collect what the audience is doing, then act on it - turned into a data habit. By 2014 the company had built TopicPulse, a content-intelligence engine that scans more than 250,000 news and social sources every night to tell a newsroom what its market will care about in the morning. It is the least flashy thing Futuri makes and arguably the most durable. A chatbot is a feature anyone can bolt on. A pipeline reading a quarter-million sources a night, tuned market by market over a decade, is a moat.

The turnWhen the tools started talking

RadioGPT stitched three things Futuri already owned into one uncanny loop. TopicPulse found the local story. A large language model wrote the script. AI voices read it out loud. Beta partners including Alpha Media in the United States and Rogers Sports & Media in Canada put it on real air. The demos were good enough that the Reuters Institute, Radio World, Inside Radio and Freethink all took a look, and good enough that the debate stopped being "can it" and started being "should it."

Then Futuri did something quietly instructive. About eight months after a launch that had earned it global headlines, it renamed RadioGPT to Futuri AudioAI. On paper that looks like a company throwing away its most famous brand. In practice it was a company reading its own runway. "RadioGPT" was a perfect name for the ChatGPT news cycle and a low ceiling for a product that was outgrowing radio into podcasts, imaging, and general audio. The lesson worth stealing: a name that maximizes attention at launch is not always the name that scales, and swapping it is cheaper than living inside the wrong box.

To broaden the voices, Futuri leaned on partners rather than building every synthetic larynx itself - integrating ElevenLabs, PlayHT and Resemble AI so a station could pick a voice that fit its format. The strategy is consistent: Futuri owns the workflow and the data, and treats the frontier models and voice engines as swappable parts. When the underlying AI improves - and it keeps improving - the product improves without a rebuild.

The real moveThen it walked over to sales

Here is the part the "AI is coming for the DJs" headlines missed. The same content engine that can draft an on-air script can draft a sales pitch, and the ad-sales desk is where the money actually hurts. So Futuri built SpotOn, which generates broadcast-ready spec spots and audio storyboards so a seller can walk into a meeting with a finished-sounding ad instead of a promise. Then it went further with TopLine, an "agentic" revenue platform - AI agents that score which accounts to chase, prep the seller before the call, and connect activity back to closed revenue.

That is the whole thesis in one motion. A media company using Futuri can find the story with TopicPulse, make the content with AudioAI and POST (which turns a live broadcast into published podcasts), package it for mobile and streaming, and sell the ads with SpotOn and TopLine. Own the find, the make, and the sell, and churn becomes a wrenching, cross-departmental decision rather than a line-item a CFO cancels on a Tuesday.

The marketWho is actually buying

Futuri sells to the people who make the noise you hear: radio and TV groups, digital publishers, agencies, and public-safety organizations. The named customer list reads like a tour of American broadcasting - iHeartMedia, Sinclair, Cumulus Media, Summit Media, Lotus, Pamal - plus Canada's Bell Media and Stingray. The company reports serving more than 7,000 brands across 22 countries, on an estimated annual revenue near $23 million, with private-equity backing from Millpond Equity Partners since 2019.

Competition comes from a few directions at once. On the AI-for-broadcast side there is Veritone and voice specialists; on the ad-sales and traffic side there are incumbents like WideOrbit and Marketron; and lurking behind all of it is the option a big group always has - stitch together general-purpose AI tools in-house and skip the vendor entirely. Futuri's counter is depth in one vertical. It is not trying to be everyone's AI. It is trying to be the media industry's AI, built by people who have spent their careers inside the format clocks and the sales books.

The catchWhere the bet gets tested

The tension inside Futuri's product is the same one inside the whole industry it serves. Broadcasting's oldest promise is that it is local - that the voice on the drive home knows your town. Automate enough of the grind and you risk automating away the thing that made local worth protecting. Futuri's answer is that the AI should handle the 3 a.m. weather read and the spec-spot first draft, freeing humans for the interviews, the calls, and the judgment that listeners actually bond with. Whether audiences experience it that way, station by station, is the real experiment - and it is being run live, on air, right now.

What a reader can take from Futuri is less about radio than about strategy. Pick an old, unglamorous industry you genuinely understand. Build the boring data pipeline first and guard it. When a technology wave arrives, do not sell the wave - sell the specific chore the wave can now erase. And when you get famous for the flashy product, quietly walk the same engine over to wherever the customer's money is really bleeding. Futuri got famous for a robot DJ. It is building a business on the sales desk.