7 Days From Payroll: How AI Saved My Startup

What we talked about.

Tony Simmons, cofounder of the podcasting and radio AI company Sonnet, joins Sunny Ray to trace his path from commercial lawyer in London and Australia to serial founder. Simmons explains how a 2016 podcasting hobby exposed him to raw transcription data packed with hidden audio characteristics, sparking the idea that became Sonnet, recently acquired. He recounts building and listing Full Circle Group on the ASX, cofounding a VR and AR company that recreated ancient cities like Jerusalem and Athens, and the harrowing early days of being seven days from missing payroll. Simmons shares how his legal training shaped how he structures fundraising, protects downside and upside, and aligns investor expectations, and how a coffee with a longtime friend led to Southern Cross Austereo becoming both investor and client. Across four companies and over $35 million raised across 22 countries, he reflects on the discipline, scars, and stubborn belief in the problem that keep pulling him back into building rather than sitting on the sidelines.

A lawyer turned four time founder on payroll scares, VR ruins, and building the AI that reshaped podcasting.

The questions, and the answers.

What does your morning routine look like these days?

It is less consistent in winter, but I get up early, clear my head, meditate, and have a coffee before meetings with my US based acquirer. Because mornings are often disjointed with calls, I push my gym session to after those meetings, training either very early or by 8 so I am done by 9 or 10. That sets up the rest of the day for productive work.

What first pulled you towards law, and was entrepreneurship always in the back of your mind?

It was more a push than a pull. My family valued professions like law, medicine, and accounting as secure, recession proof paths. I practiced for about 8 years at big law firms in Australia and with British Telecom in London, working on mergers and the rollout of Europe's broadband network. But I always wanted to do the deals myself rather than just paper them over after someone else had done them.

You made the leap from practicing M&A law to founding Full Circle, which you eventually listed on the ASX. What gave you the confidence to make that jump?

Honestly, more stupidity than confidence. I was young, 21 years ago there was barely any browser based tech, and I knew nothing about building a back end. What actually helped was starting small, working with small businesses I could manage on an Excel spreadsheet, then scaling up the technology as I won more clients. It was belief the problem was big enough and that I could solve it, not real confidence.

How did a lawyer end up building VR experiences for archaeologists?

I met my cofounder, Dr Simon Young, an archaeology PhD, at a bucks party through a friend. He was experimenting with an early Oculus Rift building ancient cityscapes, and the first version made me nearly sick with motion sickness. When the second Oculus came out with more stability, I saw the opportunity to change tourism by overlaying ancient history like the Roman Empire onto cities through VR and AR.

What was the toughest scar from your early ventures that shaped how you build companies today?

Not being able to make payroll. With Full Circle Group I put in all my savings and shares and still had to take out loans to cover gaps. Being seven days from payroll with not enough money to pay staff was the sickening moment that taught me nothing in business is a given. It created a lasting sense of urgency around governance and planning so I would never be that desperate again.

What did your legal training teach you about fundraising that most founders miss?

Structure matters. You need to understand the different ways to raise, equity, debt, safe notes, and how each protects you on the downside, but also on the upside so you are not paying full marginal tax if things go well. It also taught me that investors' expectations can differ vastly from mine, so putting a business plan and timeline on paper up front creates real alignment and protection.

When did you first realize AI could transform spoken word audio?

I started a podcast around 2016 and realized how antiquated editing was, still like cutting tape. That led me to research on the podcast industry, and eventually to a Swinburne University team whose transcription software returned over 20 hidden characteristics per word, like volume, sentiment, and pacing. I realized we could use that data to detect things like ad reads, laughter, and studio noise, and that became Sonnet.

How did you land one of Australia's biggest media companies as both investor and client?

It came down to networks and coffee chats. A longtime friend who invests in funds introduced me to another company exploring early AI, who had already been talking to Southern Cross Austereo. We all got on a call, and they invested in both companies and decided to use the technology to automate workflows across 200 radio shows and power a new app, before we had even fully gone to market.