In the last few months, Google, Meta and OpenAI have all shipped AI that works autonomously, for days at a time, without anyone watching. For B2B marketers, the interesting part isn’t the consumer launches. It’s what happens when these agents arrive at work.
Three major AI agent launches in five months
In May 2026, Google introduced Gemini Spark, a 24/7 personal AI agent, with relatively little fanfare. On 8 September, Meta launched Muse, which books travel, fills in forms and makes purchases on a user’s behalf (US-only for now). Three weeks later, on 29 September, OpenAI launched dots: always-on agents inside ChatGPT that each get their own cloud computer and connect to more than 4,000 apps.
Each of these is built to take responsibility for a job and keep working on it, rather than wait for the next prompt.
Google is bringing agents into the workplace
On 8 October 2026, at Gemini at Work, Google focused squarely on business customers with the Gemini agent. These agents get a persistent role and their own email address inside the company. Who knows, maybe a pension contribution is next.
Nick Patience at The Futurum Group made a sharper point than anything in Google’s keynote: model choice is open, but the memory, context and skills an agent builds up over time all live with the platform. I’m less interested in which model wins. What’s clear is that model choice is dissolving into agent choice.
Buying groups will soon include agents
If you sell to B2B buyers, their research will increasingly happen inside one of these agents. The agent will pick its own sources, across models, without you ever seeing it happen.
Google describes these agents as receiving “objectives, not instructions”. In practice, that might be: create a shortlist of companies that provide a particular service, for procurement to evaluate. The agent won’t have the context, preferences or relationships your sales team has built with your current buyers. It will rely on what it can find and verify.
So buying groups will probably expand to include agents. The question for every B2B brand is how you get onto an AI agent’s shortlist when it’s working on behalf of a procurement team and pulling information from other systems that decide which companies are visible.
The discipline hasn’t changed, but the audience has
My PR background tells me the fundamentals still hold. You are no longer just writing for a buyer. You are writing for whatever is vetting that buyer’s shortlist on their behalf, and that is changing fast.
The foundation for being found is still the information you put into the world, and it counts for more now, not less. That means a clear, consistent description of what you do, credible coverage from third parties, and expert content that answers the questions buyers and their agents will ask.
Would you be found by an agent working on behalf of your next customer?
SOURCES
- Google, The Gemini app becomes more agentic (Gemini Spark), May 2026
- Coursiv, Meta Muse: the personal AI agent that books and buys for you
- DataCamp, OpenAI dots: always-on agents in ChatGPT, explained
- The Futurum Group (Nick Patience), Google Gemini Agent puts the agent ahead of the model
- Layer3 Labs, Gemini agent explained
Questions
What is an AI agent?
An AI agent is an AI system that carries out tasks on someone's behalf over time, such as researching, booking, filling in forms or compiling a shortlist, rather than only answering a single question.
How will AI agents affect B2B buying?
AI agents will increasingly do early research for buying teams, including building supplier shortlists. They choose their own sources across AI models, so suppliers that aren't clearly and credibly described online risk never being considered.
How do you get on an AI agent's shortlist?
There is no guaranteed method. The foundations are the same as for AI visibility in general: a clear and consistent description of what you do, credible third-party coverage, expert content and a website that machines can read easily.