The Space Wizard
Some jobs are too big to do by hand and too one-off to ask a developer for — rename forty pages, tag every place in a region, tidy a batch of drafts. The Space Wizard is the space’s built-in assistant for exactly that: you describe what you want, and it turns the request into a plan of actions and carries it out, safely and on the record.
A request becomes a plan, then runs
The wizard works in three stages, and the order is the point:
bot interprets your request and drafts a plan of actions
assessor checks the plan — permissions, loop safety, anything destructive
executor runs the approved plan, one action at a time
Nothing happens until the plan has been drafted and assessed. You see what the wizard intends to do before it does it, and risky steps are flagged rather than run on trust. Only then does the executor carry the plan out.
It rides on the same rails as everything else
The wizard doesn’t get a private back door. Every action in its plan is one of the same operations a button or an API call would trigger — so each step is permission-checked against whoever is running it and written to the audit log just like any other change. Automation can’t do anything you couldn’t do by hand; it just does it faster and in bulk.
It works with or without AI
The wizard is always available, and it gets smarter when you give it more.
Without an AI connector, a deterministic pattern-engine interprets structured requests through fixed rules — no model, no per-use cost. This isn’t a stubbed demo: a space with no AI budget still automates real work. Activate an AI connector and a language-model client takes over the same interface, adding natural-language understanding to the same plan-assess-run pipeline. One entry point that scales with what you switch on.
Open to the space, gated on action
The wizard is a space-wide tool — everyone in the space can see its sessions and what was done. What you’re allowed to run is still governed by your permissions: seeing a plan is open, executing a destructive step is not. Visibility builds trust; the permission check keeps it safe.
The developer-facing mechanics — the pipeline internals, how an intent is assembled and assessed — sit in Automation & connectors.