Telecoms.com Podcast · Nov 9, 2026 · 1h 56m · 108m saved
With Tom Cox, Founder and CEO of Humara
Humara trained its AI sales agent on thousands of conversations in which the "AI" was really one of its best human salespeople. Tom Cox, founder and CEO of Humara, told the Telecoms.com Podcast that customers had to believe they were talking to a machine, so they would be as blunt with the salesperson as they are with a bot. Most AI agent companies chase customer service, where telecoms operators spend the most. Humara builds only for sales, which Tom calls a matter of psychology.
"We needed the end user to think that we're talking to AI, because we needed them to be as rude and as blunt as they would be without any of the social graces — to see how the human responds, because that's how we needed Humara to respond." — Tom Cox
Tom started the business 16 years ago as a chatbot that picked gifts, and turned it into a sales agent after two contacts in the industry told him the idea would work for any product.
The full episode is covered here so you can skip it. 1 hour 56 minutes of audio, 8 minutes of reading.
Here are the 9 takeaways that matter.
Key Takeaways
Humara's mission is an AI agent that sells better than the best human salesperson
The training data came from top salespeople posing as the AI, so customers would treat them like a bot
Humara builds only for sales, not customer service, which Tom sees as the bigger and more crowded market
Every message the agent writes has its own short-term sales goal, chosen from a library of sales techniques
AI shopping agents could turn brands into a checkout step, a particular risk for big telecoms operators
SpaceX's network deal with Qualcomm follows a model Viettel already runs, which Ian Fogg says should quiet the skeptics
1. 15 Gifts Came First
Tom started the company around 2010 as 15 Gifts. It was a purely conversational shopping tool, with no browsing and no product grid. Users could only talk to it.
Tom Cox's first product found gifts by reading a user's Facebook connections
"I wanted to create... the world's first ever intelligent gift-finding engine. The idea was that it would remind you every time someone important's birthday was coming up, or Valentine's Day, and it would find you 15 gifts from across the web that were perfect for that person." — Tom Cox
The engine used Facebook data to work out who a user's closest relationships were. A write-up in the Daily Mail brought a brief burst of attention, but the product never took off commercially on its own.
