What’s changed, why it matters to your work, and something useful to try next.
OpenAI is pausing further tool-use training on its most capable models and setting up a taskforce specifically to manage AI agent risk in Australia. If your organisation relies on ChatGPT's most advanced agentic tools, this signals OpenAI is deliberately slowing new agent capabilities here while it improves safeguards.
This is aimed at anyone who wants an AI agent to actually kick off work when something happens - a new email, a calendar change, a Slack message - instead of running on a fixed schedule or needing to be manually triggered each time. The Cue app is a separate early-access product for people who want a personal agent with its own contact details and a spending budget.
If you use Claude for everyday work like fixing bugs or polishing a document, Sonnet 5.5 does it faster and at lower cost than before. It is Anthropic's new "everyday" model — less powerful than the top-tier Opus 5.5, but quick and cheap enough for routine tasks.
If your team keeps re-explaining the same background context to an AI assistant separately, a shared Team Bot could hold your team's shared knowledge and files in one place so it answers consistently for anyone on the team, while still letting you have your own private conversations with it.
If you or your team use ChatGPT for work, this gives you a way to spot unauthorised access early - similar to the "recent activity" checks you might already do on Gmail or banking apps. Handy if multiple people share a ChatGPT login or you handle sensitive business info in prompts.
If your organisation uses customer-facing walkthroughs or support channels, this creates an AI avatar that can converse in real time with synchronised lip-sync and expressions across 97 languages. Available on Gemini Enterprise only, so most relevant if your org already has that plan.
If your team already uses tools like Airtable, monday.com, PandaDoc or Adobe, you can now pull them into a single Gemini conversation instead of switching between apps. For example, you could @-mention Airtable to pull project details into a chat, or bring in a PandaDoc document while drafting something else — handy if you juggle several work apps day to day.
If you regularly pull together reports or updates from scattered emails, meeting notes and files, Cowork can do the legwork for you — describe the outcome you want and it gathers the relevant information and drafts it, so you review a finished result instead of assembling it piece by piece.
If you use Microsoft Planner to track team tasks, you can now just ask it questions in plain English, like what is due this week, instead of hunting through filters and views. It is a quicker way to stay on top of project status without learning Planner's full interface.
You can now get ChatGPT to actually do work — draft a document, build a spreadsheet, or pull data from a connected app — just by talking to it, rather than typing out each instruction. Handy for hands-free moments like a commute or when you're away from your keyboard but need something drafted.
This turns Copilot into a live visual assistant — useful if you want quick answers about something in front of you, like a document, whiteboard or product, or something on your screen, without typing a description of it first. Point your camera or share your screen and just ask.
If your team uses Claude for coding, writing or day-to-day analysis, Opus 5.5 gives similar quality to the top-tier Fable 5.1 model at 40% lower cost and noticeably faster responses — worth switching to if you are already on Opus 5. It is live now across AWS, Google Cloud, Azure and Anthropic's own platform, so there is no extra setup if you already access Claude through one of those.
SCALA’s updates draw on published provider announcements. Each article separates the source summary from practical interpretation. Steve’s guides and the SCALA method help you take the next step.
Meet Steve and SCALA