Launches

RainFocus launches MCP Profiles to connect AI agents to live event data

A new RainFocus Nexus feature lets AI clients query and update live event data in plain language through OAuth-secured MCP connections.

July 8, 2026 · Confirmed with 8 sources

Cover image: RainFocus launches MCP Profiles to connect AI agents to live event data

RainFocus has launched MCP Profiles, a feature that lets AI agents connect to live event data in its platform using the Model Context Protocol, according to a PR Newswire release dated July 8, 2026 from Lindon, Utah. The feature is part of RainFocus Nexus, the company's set of AI agents for event marketing teams.

The company says teams can ask questions and make changes in plain language without exporting data or waiting on technical staff. Examples it gives include checking live session capacity, closing registration for a full track, spotting incomplete speaker profiles and following on-site check-ins.

On security, RainFocus says AI clients sign in through OAuth with per-user authentication, every action is tied to an administrator's identity, role-based controls limit what each user can do, and all activity is logged. Planners can also restrict access event by event. Setup happens in RainFocus's Integration Suite, where an admin creates a profile, switches on only the tools an AI model needs, copies the generated server URL and connects their AI client.

The release quotes RainFocus CEO JR Sherman and Marius Milcher, its VP of platform strategy and AI, along with Brent Turner of Opus Agency, a customer. The same announcement is posted in RainFocus's own newsroom.

Why it matters for event marketers: Event teams already using AI assistants can work against live registration and session data directly, with admin-level controls over what the assistant can see and change.

Scope, approval and audit: three controls on AI access to event data

When an AI agent connects to an event platform, three separate controls decide what it can do: which tools it can call at all, whether a person has to confirm before it changes something, and what record is kept afterwards. MCP Profiles handles the first through profiles. An admin creates one in the Integration Suite, switches on only the tools a task needs and gives the AI client a generated server URL. Profiles start empty, so nothing is exposed until someone enables it, Event Tech Live reported in September.

Scope is enforced by RainFocus's server. The client signs in over OAuth and receives a short-lived token tied to that profile's tool set, kept separate between development and production, so a call to any tool outside the set fails whatever the client does. Confirmation works differently. Marius Milcher told Event Tech Live that manual approval of writes is handled by the AI client rather than RainFocus's server, which he said is consistent with the current MCP specification.

The audit trail ties every tool call to the administrator whose credentials made it, and TSNN's coverage confirms per-user authentication, role-based access control and audit logging. According to Event Tech Live, though, the logs record the calls, not the prompts or the model's replies, unless a customer fills in an optional field naming the agent or skill involved.

Sorting AI tools by consequence before switching them on

RainFocus gave Event Tech Live a breakdown of its 123 admin MCP tools across 17 areas of the platform, grouped by what happens when a tool runs. By the company's count, 55 only read data, 50 create or update records in ways that can be reversed, 11 delete records or the links between them, and seven are critical because they send something out of the platform, such as an email or a badge print, or change event-level settings. Event Tech Live notes it hasn't verified the figures.

That grouping is a useful template for any admin setting up AI access. Emailing every registrant isn't a delete, but it can't be undone, so a scheme that only separates reads from writes can file it as a routine update. A safer pattern is a read-only profile for reporting and data checks, a second profile with reversible edits for a named task, and critical tools kept in a profile that only a few administrators hold.

The release's own examples are mostly read-heavy. An agent can flag attendees or speakers missing a bio, headshot or contact details, exhibitors without booth descriptions or logos, and sessions without an abstract, a room or a confirmed speaker, or compare registrations with room capacity to spot oversubscribed sessions. Opus Agency's Brent Turner said his team pulls RainFocus data into language models to build dashboards, reports and apps.

How event platforms are drawing the line on AI write access

Event platforms with MCP servers draw the line in different places. vivenu's server applies each user's own permissions to more than 100 ticketing actions, and Captello describes its CapChat server as read-only. Swoogo launched an MCP server in April 2026 and Eventtia followed in September. RainFocus's profiles add a layer on top of user permissions: the same administrator can give one AI client a narrow tool set for one job.

Real use is still early. RainFocus told Event Tech Live in September that several customers were building and testing, but none had yet run a write against a live production event. Attendee-facing tools, where the attendee rather than an admin would be the signed-in user, are on its roadmap with no committed scope or date. Milcher also said prompt injection, instructions hidden in text such as a speaker bio, isn't solved anywhere, and that RainFocus's answer is to limit what each profile can call.

What admins should ask before connecting an AI client

Ask any platform how many tools its MCP server exposes and how many of them can send data out or change event settings; whether confirmation of a write is enforced by the server or left to the AI client; and whether audit logs capture the prompt or only the call. Then agree with IT which AI clients are approved, because under this model the client is what stands between an agent and an irreversible action during a live show.

Attendee data adds its own checks: what personal data passes into the AI tool, where it's processed and whether the AI provider is on your vendor's list of subprocessors. Our guide to running a security review for event software covers what IT will want answered, and our RainFocus profile covers the rest of the platform. The release doesn't name specific AI clients or give pricing.

Events in this story

More on RainFocus

Frequently asked questions

What is RainFocus MCP Profiles?
It's a RainFocus Nexus feature, launched July 8, 2026, that connects outside AI agents to live RainFocus event data through the Model Context Protocol. An admin builds a profile with only the tools a task needs, and the AI client signs in over OAuth, with each action tied to an administrator's identity and logged.
How do you limit what an AI agent can do in an event platform?
Give it the narrowest set of tools the task needs and nothing else. In RainFocus, that means a profile with specific tools switched on, enforced by the server. Keep reporting on read-only access, put actions that send emails or change event settings in a separate profile, and use role-based permissions so the agent can't exceed its user's rights.
Who approves changes when an AI agent edits event data?
Usually the AI client, not the event platform. RainFocus told Event Tech Live that manual approval of writes happens in the client, which it says matches the current MCP specification, while its own server enforces which tools a profile can call. That makes your choice of approved AI client part of your event data controls.
Do MCP audit logs record AI prompts?
Not always. Event Tech Live reported that RainFocus logs every tool call and ties it to the administrator who made it, but doesn't capture the prompts or the model's responses. An optional field can record which agent or skill was involved. Ask any vendor what its logs capture before you rely on them for an investigation.
What is prompt injection in event software?
It's when text an AI agent reads, such as a speaker bio, session abstract or registration answer, contains instructions that try to steer the agent. Event data is full of text written by attendees and speakers, so the main defense is limiting what each agent can call: hidden text can't grant permissions the profile was never given.
When did RainFocus launch MCP Profiles?
RainFocus announced MCP Profiles on July 8, 2026, as part of RainFocus Nexus. TSNN covered it in August, and by early September RainFocus told Event Tech Live that several customers were building with it but none had yet run a write against a live production event.

Sources

We confirm every story with at least two sources before publishing. Checked 2026-11-07.

  1. RainFocus Launches AI Agent Connectivity for Event Data with MCP Profiles · PR Newswire · Press release
  2. RainFocus Launches AI Agent Connectivity for Event Data with MCP Profiles · RainFocus · Company announcement
  3. RainFocus Introduces RainFocus Nexus, a Secure, Cloud-Agnostic System of Specialized AI Agents to Act as Teammates to Event Marketers · PR Newswire · Press release
  4. Swoogo Launches Native MCP Server, Becoming the First and Only Event Platform to Connect Live Event Data to AI Tools · PR Newswire · Press release
  5. Eventtia Launches Native MCP Server to Help Event Teams Manage Events With AI · PR Newswire · Press release
  6. RainFocus: Integrations · RainFocus · Company announcement
  7. Tech Trends: New Tools for Venue Sourcing, Digital Accreditation, and More · TSNN · Press coverage
  8. The moat moved: what an event platform is worth when the interface is no longer yours · Event Tech Live · Press coverage