Getting Started with Thinking Paths
Build structured step-by-step workflows that guide learners from problem definition to completed output.
Thinking Paths vs. Prompt Assistants
A Prompt assistant answers questions. A Thinking Path does something different: it walks a learner through a structured workflow, step by step, from an initial question to a finished deliverable. You define the steps. You attach Frameworks to specific steps. The learner works through each stage in order and produces a documented output at the end.
If you want students to follow a specific analytical process (a Build vs. Buy decision, a clinical case workup, a policy brief, a design critique), you build a Thinking Path. If you want a general-purpose Q&A assistant, see Getting Started with Assistants.
Once built, a Thinking Path powers a Nudge assistant. The Nudge is what learners interact with: it runs the path, coaches them through each step, and checks their work against your criteria.
Prerequisites
You need a Course with at least one Framework created and synced. If you haven't built a framework yet, start with Getting Started with Frameworks.
The example
This walkthrough builds a Software Decision Walkthrough, a three-step path that takes a learner from problem definition to a documented Build vs. Buy recommendation. The same structure applies to any domain where you want students to move through a sequence of analytical stages.
Navigate to your course → Thinking Paths tab → New Thinking Path. The form opens with four tabs: Overview, Guardrails, Final Output, and Tutoring Approach. Fill them in order.
Overview tab
Name
Give the path a clear, specific name.
Example: Software Decision Walkthrough
Purpose
Describe the output the learner will produce by completing the path.
Example: Guide a learner through the full Build vs. Buy analysis, from defining the problem to producing a documented recommendation with justification.
Guardrails tab
Guardrails are rules the assistant enforces throughout the workflow. If a learner tries to skip ahead or submit incomplete work, the assistant redirects them based on these rules.
Rules
Click Add Rule and write each rule as a direct instruction.
Example:
- Rule 1:
Do not move to the recommendation step until the learner has evaluated both the build cost and the buy cost.
Reasoning Questions
Reasoning Questions are internal prompts the assistant processes before responding. For example, a reasoning question like Has the learner addressed all three cost categories? causes the assistant to verify coverage before giving feedback, rather than accepting a partial answer. You can add these later once you've tested the path.
Final Output tab
Describe what a complete, high-quality deliverable looks like. The assistant checks the learner's submission against this description.
Example: A Build vs. Buy recommendation document containing: (1) a clear statement of the problem the software must solve, (2) an evaluation of internal build capability, (3) a cost comparison, and (4) a final recommendation (Build or Buy) with a one-paragraph justification.
Tutoring Approach tab
Set the coaching style for the assistant.
Example: Ask Socratic questions rather than giving answers. When a learner is stuck, point to the relevant framework input, not the answer. Acknowledge completed steps briefly, then redirect to the next one.
Click Create Path.
Create steps
Open your saved path from the Thinking Paths tab and click Add Step. The Software Decision Walkthrough has three steps:
Step 1: Define the problem
| Field | Value |
|---|---|
| Name | Define the problem |
| Instructions | Articulate the specific problem the software needs to solve. What does success look like? What constraints exist (budget, timeline, team size)? |
| Checklist | The learner has written a clear problem statement with success criteria. |
Step 2: Evaluate the options
| Field | Value |
|---|---|
| Name | Evaluate the options |
| Instructions | Apply the Build vs. Buy framework. Assess internal capability, estimate build cost, and identify available products with their costs and fit. |
| Related Frameworks | Select Build vs. Buy Software Decision |
| Checklist | The learner has assessed internal capability, estimated build cost, and identified at least one buy option with a cost estimate. |
Step 3: Make a recommendation
| Field | Value |
|---|---|
| Name | Make a recommendation |
| Instructions | State your recommendation: Build or Buy. Justify it in one paragraph covering capability, cost, and strategic fit. |
| Checklist | The learner has produced a written recommendation (Build or Buy) with a justification paragraph. |
Click Create Step after each one.
Connect steps in the graph view
After creating all three steps, each appears as a node on the canvas.
- Drag from Define the problem to Evaluate the options.
- Drag from Evaluate the options to Make a recommendation.
- Open the path's Overview tab and set Entry Step to Define the problem.
Deploy as a Nudge
To make the path available to learners, attach it to an assistant.
- Navigate to your course → Assistants tab → Create Assistant
- Choose Thinking Path as the assistant type
- Select Software Decision Walkthrough
- Name it (e.g.,
Build vs. Buy Advisor) - Configure Access Level, Conversation Starters, and AI Model
- Click Deploy
The assistant appears to learners as a Nudge. It walks them through each step, coaches them using your tutoring approach, and checks their final output against your criteria.
Update or delete
To edit the path, its steps, or to delete it, see Update a Thinking Path.
Learn more
For path-level fields and the graph view, see Thinking Paths Reference. For step-level fields, see Steps Reference. To understand the learner experience and Nudge configuration, see Use a Thinking Path.