What’s changed, why it matters to your work, and something useful to try next.
For people running regular automated tasks — a weekly data pull or status update — everything now stays in one ongoing conversation with a calendar view, so you can see and adjust the whole run history in one place instead of digging through separate task threads.
Useful for anyone who finds themselves re-explaining the same preferences to Grok every session — a preferred report format, a tone of voice, a recurring checklist. Skills let you teach it once and have that workflow remembered and reused automatically in future conversations, cutting out repetitive setup instructions.
Useful for anyone doing everyday admin on their phone: you can describe a multi-step task in plain language and let Android carry it out, or have Chrome summarise and act on a webpage instead of reading it end to end. It also lets non-developers build a custom home-screen widget just by describing what they want.
For Mac users, Perplexity's assistant can now securely use your local files, apps and browser as context when answering questions, with an upgraded "Personal CFO" feature too. Worth trying if you want AI help pulling together your own documents or personal finances without manually uploading files each time.
For anyone running the same kind of project repeatedly in Manus (recurring reports, ongoing research), it now suggests refinements to its own setup based on what worked in past tasks. You just approve or reject each suggestion rather than manually retuning instructions every time.
This is an infrastructure and business deal between xAI and Anthropic over compute supply, not a product change — there is nothing here that changes what you can do day to day. Relevant mainly as a signal of the AI industry's underlying compute economics, not a tool to use.
If your job involves juggling SharePoint, Outlook, Google Workspace, Notion or GitHub, this lets Grok read and act on that data directly instead of you copying and pasting content into a chat window. Practically, you could ask it to pull details from an email thread, a Notion doc and a spreadsheet in one go and get a combined answer, saving the manual stitching-together work.
Squarely for finance professionals — analysts and research staff can now pull licensed market and company data from providers like Morningstar and FactSet straight into their AI research sessions, instead of switching between separate paid terminals.
For anyone using Perplexity to look up health questions, answers now draw on clinical-grade sources like the New England Journal of Medicine and BMJ, with citations shown. Useful for getting better-grounded information before discussing something with a doctor, though it's still not a diagnosis.
If you work in finance, accounting, or a related back-office role, these are pre-built starting points rather than a from-scratch AI project — ten ready-made agent setups for common financial-services tasks that plug into the tools your team may already use (Claude Code, Managed Agents) plus new add-ins for Word, Excel and Outlook. Worth a look if you have been wanting to automate routine finance workflows but did not want to build the AI plumbing yourself.
Mainly relevant to organisations already on Perplexity's paid business plans — the underlying AI model and image generator got upgrades, and teams using Microsoft Teams or data platforms like Snowflake and Databricks can now connect Computer directly to those tools, plus publish simple websites from within it.
Worth turning on if your ChatGPT account holds business-sensitive conversations or admin access — this optional setting lets you require a stronger sign-in method (like a passkey or hardware security key) instead of relying on a password alone.
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