This Week Three Companies Shipped Agents That Do Your Tasks

Three companies shipped AI agents this week that do concrete tasks instead of just answering questions. OpenAI released GPT-6 Astra, Salesforce made six Agentforce agents generally available, and Meta launched a personal agent called Muse. If your job involves repeatable tasks, at least one of these three now reaches into it directly.

What did AI ship this week?

Three separate companies moved AI from chat into action within the same week of September 2026. OpenAI unveiled GPT-6 Astra on September 4, 2026, calling it its most capable model yet, according to Al Jazeera’s coverage of the release. Salesforce announced on September 11, 2026 that six of its seven named Agentforce agents were generally available, per Salesforce’s own announcement. Meta introduced Muse, a personal AI agent, in the United States on September 8, 2026, as detailed in Meta’s launch post. Each product is built to plan and carry out several steps of a task on its own, not just answer a single question, which changes what having AI at work looks like for the rest of this year.

Is GPT-6 Astra different from past releases?

Yes, mainly because of where it sits on OpenAI’s own risk scale, not just what it can do. GPT-6 Astra is state of the art on coding, computer use and professional tasks, according to Al Jazeera’s report on the release, which also notes it is the first OpenAI model gated at the company’s highest cybersecurity risk tier. That gating points to a slower, more restricted rollout to individual accounts than past releases got, per the same report. If your team has been waiting to test a frontier model on everyday workflows, expect access to arrive in stages. The computer use capability is worth watching closest, since it lets the model operate software the way a person would, a concept covered in more depth in Long Context, Tool Use And Computer Use Explained Plainly.

What can Agentforce’s seven agents do?

Salesforce built seven named agents, each assigned to a specific job function, and put six of them into general availability on September 11, 2026. According to Salesforce’s announcement, the agents cover service, IT and HR, ecommerce, supply chain and sales, and are designed to run full workflows inside tools sales and operations professionals already use, not answer questions in a chat window. Salesforce calls them “job-ready” agents, which is a vendor description of intended capability, not an independent finding. If you work in sales, service or supply chain, this is the release to watch closest, because it positions these agents as coworkers assigned specific duties, distinct from a chat assistant answering one question at a time.

Should you try Meta’s Muse this week?

Yes, because it costs nothing to test at the basic tier and takes only a few minutes to set up. Meta launched Muse in the US on September 8, 2026 as a personal AI agent that plans and carries out tasks such as booking travel and sending emails, reachable through WhatsApp or its own app, according to Meta’s announcement. It is free for basic use, per the same post. Before connecting it to email or a calendar, check exactly what the agent needs access to in order to complete a task, since letting an agent send messages on your behalf is a permission decision, not a convenience toggle.

Does the 49% layoff stat prove AI is the cause?

No, and this is where the data needs a careful read. Skillsyncer’s layoffs tracker reported on September 17, 2026 that 49% of 2026 layoffs cite AI or automation as a factor, with programmers, customer service, data entry and QA roles named most often in the same tracker. A layoff that names AI as a factor is not proof that AI was the sole or even primary cause. A company naming automation in a layoff announcement may be describing a genuine substitution, cost cutting framed as modernization, or a restructuring driven by something else entirely. The tracker also does not claim to cover every layoff worldwide, so the 49% figure describes what it captured, not the full labor market. For a longer look at how this trend built across the year, see What Half Of 2026’s AI Layoffs Mean For Your Job.

Four roles most exposed to these agents, ranked

Not every job sits equally close to what shipped this week. Ranking these by how directly a named product now touches daily tasks in that role shows where to pay attention first.

1. Customer service. Agentforce’s service agent and Meta Muse both automate direct handling of customer or personal requests, and Skillsyncer’s tracker names customer service among the roles cited most often in AI-linked layoffs.
2. Coding and QA. GPT-6 Astra is state of the art on coding tasks, according to Al Jazeera, and programmers and QA roles appear most often in the same layoffs data.
3. IT, HR and back-office admin. Salesforce’s IT/HR agent is built to run full internal workflows, according to Salesforce, putting repeatable admin tasks directly in scope.
4. Sales and ecommerce operations. Agentforce’s sales and ecommerce agents handle workflow steps that used to need a person at each stage, per Salesforce’s release, though this remains a vendor description of intended use rather than a measured outcome.

What should you do about it this week?

Test one of these tools directly instead of only reading about it. Meta Muse is free to try today, and Meta’s own setup materials walk through the basics. If your company uses Salesforce, ask your manager for the specific Agentforce rollout timeline instead of guessing from the announcement. If your function was named in Skillsyncer’s tracker, treat AI fluency as a job requirement now, not a future skill. If you need help explaining AI-assisted work to a manager who does not trust it yet, this guide covers how to do that without overselling what the tool did.

You are a career strategist helping me figure out which parts of my job could realistically be done by an AI agent within the next 12 months. My job title is [insert title] and here is a list of my main weekly tasks: [paste 8 to 12 tasks]. For each task, tell me whether a general-purpose agent like Meta's Muse or a workflow-specific tool like Salesforce Agentforce could plausibly handle it today, partially handle it, or not touch it at all, and explain why in one sentence per task. Then tell me which 2 tasks I should get hands-on AI experience with first, and why those two. Ask me clarifying questions about my tools or workflow before answering if you need them.

Help me write 5 direct questions to ask my manager about how our company plans to roll out AI agents like Salesforce Agentforce or similar tools in the next 6 to 12 months. I want questions that get specific answers about timeline, which of my tasks might change, and what training or support will be offered, not vague reassurance. My role is [insert role] and my team currently uses [insert tools, e.g. Salesforce, internal ticketing system]. Keep the tone professional and curious, not anxious or accusatory. Ask me for more context about my company or team if you need it before drafting.

I am about to try Meta's Muse personal AI agent for the first time to help with everyday tasks like scheduling and email drafts. Before I connect any accounts, walk me through a checklist of what permissions I should check first, what kind of task I should test with low risk before trying anything sensitive, and what warning signs would tell me to disconnect it. Assume I have no prior experience granting an AI agent access to my email or messaging apps. Ask me what accounts I am considering connecting before giving your final checklist.

Act as a blunt hiring manager who has read ten thousand resumes, not a career coach. I will paste my full resume and the job description I want. Rewrite the whole resume for that role, section by section, in this order: summary, experience, skills, education. Rules: every experience line leads with impact, not duty. Use bracketed placeholders like [8 percent] for any number I did not give you, and list at the end every placeholder I need to replace with a real figure. Keep it to one page of text. Plain formatting only, no tables or columns, so screening software can parse it. After the rewrite, tell me the three weakest claims that need evidence before I send this anywhere. My resume: [paste resume]. The role: [paste job description].