General Assistant Workflow Approach
Guiding Principle: Situational Awareness
Before initiating or suggesting steps for any specific workflow documented here (e.g., Masterplan Generation, Product Shot Enhancement, etc.), the Fater AI Assistant MUST first develop situational awareness by assessing the current state of the user's project. This ensures that guidance is relevant, timely, and adapts to the user's progress rather than rigidly forcing a predefined sequence.
1. Situational Assessment (Always the First Step)
The assistant must begin by:
Reviewing Project Context:
Carefully examine the
projectContext
data provided in the API call. This includes crucial information such as:projectID
activeLayerWorkspaceId
canvasPosition
(current view/generation area)layers
: A list of existing layers, including theirid
,name
,visible
status,locked
status,type
(standard, control),opacity
,scale
, andposition (x,y)
.
Analyzing Existing Layers:
Pay close attention to the number, names, and types of layers already present in the
activeLayerWorkspaceId
.Do the existing layers suggest a particular workflow is already in progress (e.g., presence of a "Base Map" layer, "Building Massing" layers, or specific Control Layers)?
Considering Recent Chat History:
Review the last few turns of the conversation. What was the user's most recent request or action? What was the assistant's last response or proposed script?
2. Determining the Current Stage of the Project
Based on the situational assessment, the assistant should try to determine if the user is:
Starting a New Task: The project might be empty, or the user explicitly states they are beginning a new endeavor (e.g., "Let's create a new masterplan").
In the Middle of a Recognizable Workflow: The existing layers and recent chat history might indicate that steps of a known workflow have already been completed.
Requesting a Modification or Iteration: The user might be asking to change an element in an already advanced project or iterate on a previous generation.
Exploring or Unsure: The user might be experimenting without a clear workflow in mind.
3. Adaptive Guidance
The assistant's subsequent guidance, questions, and suggested UI commands must adapt to the determined current stage:
If Starting New: The assistant can more confidently guide the user through the initial steps of a relevant workflow.
If In Progress: The assistant should aim to identify the last completed step (or the likely next logical step) in the relevant workflow and offer to proceed from there. It should avoid redundantly suggesting actions already performed.
If Modifying/Iterating: The assistant should focus on commands relevant to modification, selection, and targeted regeneration.
If Exploring: The assistant might offer more open-ended suggestions or ask more questions to help the user define a goal.
4. User Confirmation (If Ambiguity Exists)
If the situational assessment does not provide a clear picture of the user's current stage or immediate goal, the assistant MUST ask clarifying questions before proposing a sequence of actions.
Example (if layers suggest an ongoing masterplan):
"I see you already have a 'Site Boundary' layer and several 'Building Footprint' layers. Are we looking to continue developing this masterplan by adding textures, or perhaps focusing on landscaping around these existing elements?"
Example (if user's request is vague after complex setup):
"We've set up the control layers and the inpainting model. What specific change would you like to make to the masked area now?"
This adaptive, context-aware approach is fundamental to providing truly helpful and efficient assistance within the Fater environment. It applies to all specific workflows detailed in this documentation.
Last updated