Building your own AI sounds like the smart, cost-effective move. More control, no vendor lock-in, total flexibility. But according to Jimmy Hosang, a former data scientist who spent years doing exactly that, the reality is far messier and potentially damaging to your customers’ trust.
In this CX Today conversation, Jimmy brings a rare perspective. He’s the poacher-turned-gamekeeper who once turned vendors away to build everything himself, and now runs Mojo CX, an AI platform for contact centers. That background makes him uniquely well-placed to cut through the noise.
And there’s plenty of noise to cut through. Jimmy argues that most CX leaders are making the same mistake with AI that they made with chatbots and digital transformation: buying (or building) before they know what problem they’re actually solving.
The result? Hidden costs stack up, failure rates are sky-high, and the shiniest use cases, like voice AI, turn out to be the riskiest and most expensive places to start.
Instead, Jimmy lays out a clear maturity path: start with transcription and auto-summarization, move into auto QA and agent coaching, then agent assist, and only reach for voice AI once you’ve built the data foundation to do it safely.
He also makes a compelling case for thinking of AI agents and human agents as one team, and explains why AI feedback might actually land better with frontline staff than feedback from their own managers.
If you’re trying to figure out where to start, or where you’re going wrong, this one is worth your time.
