When the agent RSVPs
Personal AI agents can now fill out forms, send invitations and make calls on people's behalf. For anyone who runs a guest list, the RSVP is about to mean less, and the check-in at the door is about to mean more.

This month, two of the largest consumer technology companies started handing people's errands to software.
Meta announced Muse on September 8 and described it as a personal agent that "can open a browser, fill out forms, and negotiate on their behalf" (Meta, 2026). Apptopia estimates it passed 2.8 million installs in its first 12 days (TechCrunch, 2026). On September 24, Google began testing "Call for Me," which lets Gemini phone a business for you to make a reservation or move an appointment, using your own number (TechCrunch, 2026).
The friction showed up within days. Amazon blocked Muse from its site, saying the agent did not identify itself as an agent in its requests, as Amazon's conditions of use require (GeekWire, 2026). Box CEO Aaron Levie told CNBC that the industry is "basically two weeks into this entire topic" (GeekWire, 2026).
Most of the coverage is about shopping. The same question is coming for anyone who sends invitations: what does a yes mean when software can say it?
The RSVP was already soft
Event teams have never fully trusted a registration. They just knew roughly how much to discount it.
Clutch Events, a vendor that runs executive events, tracked 191 events and more than 13,000 registrations. Attendance ran 70% for executive dinners, 61% for private roundtables, 52% for networking dinners and 19% for vendor-owned conferences (Clutch Events, 2026). Across more than 860 events, PheedLoop found a median no-show rate of about 28% for free events (PheedLoop, 2026). Both are vendor datasets, but the lesson is familiar: a registration is a statement of intent, and intent decays.
What held the system together was effort. Registering took a few minutes of someone's attention, and that small cost filtered out some of the people who were never coming.
What changes when software answers
Nobody has published data yet on how agents behave with event invitations, and it would be wrong to pretend otherwise. But the capabilities are public, and a few effects are reasonable to expect.
Saying yes gets cheaper. An agent that can fill out forms can fill out a registration form in seconds. When the cost of saying yes drops, registration counts tend to rise faster than attendance.
The registrant may not be the decider. Meta says Muse checks with the person before sensitive actions like sending an email or making a purchase (Meta, 2026). Whether a free event registration counts as sensitive will depend on how each assistant is built and set up. Some yeses will be considered. Some will be delegated.
Your data may not show the difference. Amazon's complaint was that it could not tell the agent from the customer. A registration form has the same problem. An agent-completed form looks exactly like a human one unless the agent announces itself, and most registration forms have no way to ask.
Declines and reschedules get easier too. This is the good news. A person who would have quietly skipped your dinner may now have an assistant that cancels for them, which gives you a seat back in time to fill it. In the Clutch data, roundtables already see about 11 advance cancellations for every day-of no-show (Clutch Events, 2026). Cheap cancellation could make that ratio better.
None of this is fully here yet. Calendar tools are still limited: Gemini's scheduling assistant in Gmail, for example, was designed for one-on-one meetings, not group invitations (TechCrunch, 2025). But the direction is clear, and 2027 event calendars are being planned now.
The planner's side is automating too
The same shift is happening on the organizer's side of the list. At Skift Meetings Forum this week, Vishal Joshi, co-founder and CEO of Joy, said his company's AI makes thousands of calls a day to hotel front desks to confirm reservations, and described a 200-person CIO event planned by two people with AI research support (Skift Meetings, 2026).
Gartner predicts that by 2028, agentic AI will make at least 15% of day-to-day work decisions autonomously (Gartner, 2025). Invitations, confirmations and reminders are exactly the kind of routine back-and-forth that gets handed off first. Before long, some of your RSVP traffic may be one piece of software replying to another.
What to change before 2027
Executive events already have a model for this. Senior guests have long had assistants accept invitations for them, and good event teams learned not to confuse the assistant's yes with the executive's. An AI agent is that assistant for everyone. The playbook is similar.
Confirm with the person, close to the date. A reconfirmation 48 to 72 hours out, sent to the guest directly, separates considered yeses from delegated ones. Ask something only the guest can answer: a dietary preference, a colleague they want to bring, a topic they want covered.
Treat registration as a forecast. Report registrations, but plan catering, seating and follow-up around your historical show rate by event type and seniority. If the ratio between the two starts to drift, that is your early signal that delegated yeses are arriving.
Make the door the record. Check-in is the one step no agent can do for your guest. The attended list, captured at the door, is the only list that says who was actually in the room. That list, not the registration export, is what should flow into the CRM and next to open opportunities.
Watch who is coming back. A guest who attended your last two dinners is a stronger signal than any form. Attendance history cannot be delegated, and it becomes more valuable as registrations become less reliable.
That last point is where Socially sits. It runs check-in at the door, syncs actual attendance with HubSpot or Salesforce, and shows which open opportunities had contacts in the room, so the record the team reports on is who showed up, not who clicked yes.
A yes is cheap. Being there is not.
For years, the registration count was a decent proxy for demand, because registering took a little effort. That effort is going away. The teams that plan for it will stop treating the RSVP list as their scoreboard and start treating the room as the only number that counts.
Agents can accept invitations, but they can't show up for your guests. The event record should be built around the part that still takes a person.
Sources
- Meta, Introducing Muse: The World's First Personal AI Agent Built for Everyone, 2026: https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/
- TechCrunch, Meta's Muse is outpacing ChatGPT's early mobile launch, 2026: https://techcrunch.com/2026/09/21/metas-muse-is-outpacing-chatgpts-early-mobile-launch/
- TechCrunch, Google tests letting Gemini call businesses for you, 2026: https://techcrunch.com/2026/09/24/google-tests-letting-gemini-make-phone-calls-initially-for-us-pixel-owners/
- GeekWire, Amazon's fight with Meta: Who owns the customer relationship when an agent does the buying?, 2026: https://www.geekwire.com/2026/amazons-fight-with-meta-who-owns-the-customer-relationship-when-an-agent-does-the-buying/
- Clutch Events, The State of Executive B2B Event Attendance, 2026: https://report.clutchevents.co/
- PheedLoop, Event Data Lab Report #06, 2026: https://web.pheedloop.com/blog/event-data-lab-report-06
- TechCrunch, Google's Gemini can now help you schedule Google Calendar meetings, 2025: https://techcrunch.com/2025/10/14/googles-gemini-can-now-help-you-schedule-google-calendar-meetings/
- Skift Meetings, What Autonomous AI Means for Event Planning, 2026: https://meetings.skift.com/2026/09/23/what-autonomous-ai-means-for-event-planning/
- Gartner, 3 Bold and Actionable Predictions for the Future of GenAI, 2025: https://www.gartner.com/en/articles/3-bold-and-actionable-predictions-for-the-future-of-genai
Newsletter
The High Signal
A weekly read on AI, events and go-to-market, from Jeremy Vargas and The Guest List. Free, every week.
Unsubscribe from any issue. Privacy policy.


