More Control, Less Noise: AI Updates for Developers, Creators, and Freelancers

If this week's AI news has a theme, it's control: over how much reasoning your coding assistant spends, over how a video draft gets built, over whether your best prompts get reused or retyped, and over whether your wrist buzzes or stays quiet. GitHub Copilot tightens its Visual Studio workflow, Gemini Omni 1.1 Flash opens up video drafting, Mozilla's Octonous turns prompts into reusable skills, and Wired argues the most interesting wearables are the ones that leave you alone.

GitHub Copilot in Visual Studio: reasoning dials and pre-commit AI reviews

GitHub's August update to Copilot in Visual Studio adds two controls aimed squarely at everyday coding. The first is a set of reasoning dials: you can adjust how much reasoning effort Copilot puts into a request, trading depth against speed depending on whether you want a quick suggestion or a carefully worked answer. The second is AI review before you commit: changes get a Copilot pass before they land in your history, catching problems while they are still cheap to fix. Why it matters: both features shift the trade-off decisions from default settings to the developer. If you work in Visual Studio, update the extension and try the review step on your next commit — a low-risk way to see whether an extra AI pass fits your flow. [S6]

Gemini Omni 1.1 Flash: more control, and video drafts at 360p

Google's developer blog presents Gemini Omni 1.1 Flash as a way to build with more control, and one detail stands out for creators: the model can draft video at 360p. That makes a fast first pass practical — sketch a visual concept, judge whether the direction works, and only then invest in something more polished. Why it matters: iteration speed, not final resolution, is usually the bottleneck in creative work, and a low-friction draft stage changes how many ideas you can afford to test. How to respond: if you build with Gemini, Google's post is worth reading for the control options it highlights, which are aimed at developers who want to shape outputs rather than accept whatever arrives. [S9]

Octonous Agent Skills: your prompts, reused

Mozilla AI's Octonous introduces Agent Skills, built around a simple idea: the prompts you have already refined should not be retyped. Instead of reconstructing the same instructions — tone, constraints, output format — every time you start a task, you save them once as a skill and reuse them across agent work. Why it matters: prompt quality is one of the biggest levers individuals have over AI output, and reuse turns a one-off good result into a repeatable one. How to respond: if you use Octonous, start with the prompts you rewrite most often; those are the strongest candidates to become skills. [S11]

Wearables that want to be ignored

Wired surveys the hottest new wearables and lands on a counterintuitive design goal: they want to be ignored. Rather than competing for your attention with a stream of alerts, this generation leans on quiet operation and daily summaries — devices that batch what happened and let you check in on your own schedule. Why it matters: for freelancers and knowledge workers, an always-on notifier is a direct tax on deep work, and a summary-first wearable inverts that relationship. How to respond: if notification fatigue is real for you, it may be worth judging wearables by how little they interrupt you, not by how much they do — the distinction Wired draws is becoming a genuine buying criterion. [S4]

What to watch next

The common thread is control applied in the right places: more of it where precision matters, in Copilot's reasoning dials and Gemini's build options; less friction where speed matters, in Octonous's reusable skills and 360p video drafts; and less noise where attention is the scarce resource, in summary-first wearables. None of these demand an enterprise decision — they are one update, one saved prompt, or one different question at checkout away.