Skip to content

The project assistant

Ask a project questions from Rhino or the web. Answers are built from the project's own data, and you can see what each one was built from.

ProGuide · 3 min read

Every project has a chat. In Rhino it fills the Projects view of the Wrengler panel; on the web it is the Chat tab of the project page. Both show the same conversations.

This page is about asking. In Rhino the assistant can also act on the open model and its project: see The assistant in Rhino.

Point a question at something#

Press + in the message box and pick what the message is for:

ChoiceThe assistant readsIn RhinoOn the web
Project datathe project's facts and tags, its documents, the conversation and your saved notesYes, plus a short summary of the open fileYes
Rhino scenea short summary of the open file (units, site, selection and its areas, groups), the project's facts, the conversation and your notes; not the documentsYesNo
Imagethe conversation onlyYes: it proposes an image, from the Rhino view or a description, and the Image Studio draws itYes: from a description, drawn in the chat
Goalnot a question: agents work through the project and post a cited answerYesYes
What you pick decides what the assistant can read. Rhino scene adds a summary of the open file, never its geometry.

With nothing picked, the message is about Project data. In Rhino the chat needs the open file to be linked to a project before it can send.

For something bigger than one question, pick Goal. See Goals.

Numbers come from your data#

When the assistant states a number (a floor area, a sun-hours result, a count), it quotes a value stored in the project, as stored. If the project has no such value, it says so rather than estimating one. Tag the facts you care about (gfa = 12000 m²) and they become answerable.

The assistant quotes numbers from what the project stores and names the source.

What an answer was built from#

Beside Send, a small ring fills as the conversation uses its room. Press it to open Conversation details:

  • Context: how full the conversation is.
  • Built from: how much of each source the last answer read. The sources are Instructions, Project facts, Rhino scene, Documents, Conversation and Memory. If something was left out for room, it says what.
  • Summarised: how many earlier messages have been folded into a summary.
  • Last answer: the model, how deep it went (Quick, Light or Deep), and how long it took.

The ring appears once an answer has been given. An instant command in Rhino, such as select the podium, reads nothing and shows none.

What it offers to save#

The assistant never saves on its own. When an answer contains something worth keeping, it shows a card, and nothing is saved until you press the card's button:

CardButtonWhat it saves
A tagSave tagOne fact on the project, such as typology = courtyard
A noteRememberA note for you, or for the project, that later answers can read
A document (web)Save documentA study's result, written into the project's files
An imageGenerateNothing yet: it draws the image. On the web the image appears in the chat, with Save to project, Download and Prompt

An image drawn in the web chat is not kept in the conversation. Save it to the project, or download it, before you leave the page.

Privacy#

  • The assistant only sees projects you can see. Every message runs as you, under the same access rules as the rest of your account.
  • What it gets of your Rhino file is a short factual summary, not your geometry. Your model never leaves your machine.
  • Conversations are saved with the project and deleted with it, or with your account.
  • An organization can switch AI features off for its projects. The chat then says so.

Plans#

The assistant comes with Wrengler Pro, in Rhino and on the web. See pricing.

Tip

Nothing to ask yet? Add a few tags to the project on the web (typology, client, floor area), then ask the assistant to compare or summarize them.