Token prices dropped 50x in two years. Enterprise AI bills keep climbing anyway.
Jevons paradox. Cheaper tokens mean more usage, not lower spend.
Waiting for prices to fall is not a cost strategy.
Instinct is hot. But Optimizely is hotter for the enterprise. We just launched our version of AI assistants, except they are agentic marketers and ready to work alongside your human marketing team. We have them in place at dozens of customers. optimizely.com/solutions/vir…
"Employees on day one are given a computer, credentials, and access to the systems and information they'll need to do their job....Things are no different with virtual teammates."
Our SVP of Product Kevin Li in @FastCompany by @BrettFarmiloe: optimize.ly/4rC2W7h
We tested Jev against our production skill selection process this weekend. The results pretty stunning. Speed: Jev was faster at every catalog size, by 15% to 36%. Cost: 7x to 20x cheaper.
When I read the Jev announcement I thought it was just a better and faster LLM. But...
We're post-training our own models for marketing. And they're beating frontier models on quality — at 10x lower cost.
A few things we announced at Opticon:
1) Purpose-built marketing models. We announced a family of post-trained models tuned specifically for marketing tasks.
Awesome few days with the @Optimizely community at Opticon 2026 in New York. 🏕️
There was a lot to take in. Alexander Atzberger talked about how AI can unlock talent and help people achieve more.
And Lauren Hammarstedt gave us a look at what’s next with Virtual Teammates, designed to proactively work alongside teams with their own identities, permissions, and responsibilities.
Plenty to think about, and even more to put into practice for our clients.
Thanks to everyone who stopped by to meet XCentium and to the entire Optimizely team for an amazing event and a great lineup of speakers.
👉 Learn more about XCentium’s Optimizely capabilities: xcentium.com/platforms/optim…#Opticon2026#Optimizely#AI#DigitalExperience