What the agentic harness is
The agentic harness lets a trusted AI assistant carry out real tasks in a venue’s system, not just talk about them, always inside the venue’s own rules and permissions.
When you'd use it
Read this first to understand what the AI runs on and the guardrails around it. This is the foundation layer; Agent Chat is where you use it.
How it behaves
The agentic harness is the foundation the AI runs on: a connected graph of your guest data, a real toolset the agent can use, and the control to run it safely. It is the plumbing that makes an agent useful rather than a chatbot that guesses.
The graph joins every guest to their bookings, spend, Wi-Fi, reviews and loyalty by the relationships between them, including real relationships like who dined with whom and who referred whom, plus computed signals like churn risk and lifetime-value tiers. A question that would take nine database queries becomes a single traversal.
The toolset lets the agent query the graph and your data, browse the web, and build a report or dashboard as a shareable artifact. It reads and drafts; anything that would change your data waits for your sign-off, and access to the underlying data is read-only.
Configuration and options
- The graph
- Guest data joined by relationships, with computed signals like churn risk and lifetime value.
- Toolset
- Query the graph and data, browse the web, build shareable artifacts.
- Read-and-draft
- Reads and drafts; any change to data waits for sign-off.
- Where it is used
- Through Agent Chat, in plain English.
Good to know
This is a foundations-first approach: the connected data and toolset came first, and the agent capabilities are early and growing on top.
It is your own data on your own account; the agent is kept from revealing secrets or internal config.