AI Pulse · AI Trends Pulse

Google has released two versions of Gemini 3.8 aimed at different kinds of work. Gemini 3.8 Live is built for real-time interaction. Gemini 3.8 Live Extended Thinking is the separate option for tasks that need more deliberate reasoning. The key point is not that one model replaces the other. Google is making the trade-off explicit, fast responses when speed matters, deeper processing when the job is more complex. You can read the details in Google’s announcement.

For a business owner, this is a useful way to think about AI deployment. A customer-facing assistant, internal help desk, meeting companion, or sales call tool usually needs to respond quickly. A model reviewing a contract, analysing a board pack, checking a forecast, or working through a complex operational issue may need more time to produce a considered answer. Treating both jobs as if they need the same AI model can mean paying too much, waiting too long, or getting a shallow result on work that needs scrutiny.

The early Hacker News discussion reached 74 points within hours, which suggests developers are paying attention, though that is not a measure of business value. The practical question is whether your team can route work to the right type of model and set clear rules for when a slower, deeper answer is worth it. This is the kind of thing we build into an AI command centre, so fast operational requests and higher-stakes reasoning work do not get handled the same way.