All use cases
Education

Guided learning assistant

Help learners think independently with scaffolded AI support

Most AI tools give students instant answers, which can undermine learning. A guided learning assistant takes a different approach. Grounded in established pedagogical methods including Socratic questioning, scaffolded instruction and metacognitive reflection, it asks what the learner already knows, identifies confusion points and provides hints gradually. If a student genuinely struggles after multiple attempts, the AI increases its support level from reflective questions to guided hints to structured explanations. Teachers can monitor engagement, see which topics students find hardest and generate question sets for their lessons. The system encourages reflection and independent thinking rather than dependency.

How it works

Here's what a real conversation with your ThinkGuide assistant looks like.

ThinkGuide

AI assistant

I don't understand photosynthesis
No problem, before we break it down, what part feels most confusing to you? For example, do you know what plants need to grow, or is it the chemical process that's tricky? Let's start from what you already know and build from there.

What you get

Scaffolded support with four progressive assistance levels

Guided, Explain, Revision and Staff Assist modes

Learner reflection prompts and confidence tracking

Teacher dashboard with topic heatmaps and engagement insights

AI guardrails to prevent instant answer generation

The outcome

Learners develop stronger independent thinking skills. Teachers gain visibility into where students struggle most. The college introduces AI in a way that supports learning quality rather than shortcuts.

Set it up in 6 steps

1

Create a ThinkGuide assistant from the college template library

2

Upload course materials and study guides as knowledge sources

3

Configure AI guardrails (answer thresholds, reflection requirements)

4

Set up learner profiles with accessibility preferences

5

Deploy on your student portal or VLE via the web widget

6

Monitor the Guided Learning Dashboard for engagement insights

How each feature helps

These are the platform features that make this use case work. Each one plays a specific role in delivering results.

Knowledge sources

You give your assistant information to learn from. Upload documents, paste URLs or connect cloud folders. The assistant reads everything, breaks it into chunks and uses the relevant bits when answering questions.

Actions

Actions let your assistant do things in the real world. Send an email, create a support ticket, update a CRM record or call any webhook. You decide which actions need human approval before they run.

Human handoff

When the assistant can't help or the customer asks for a person, it hands the conversation over to your team. The handoff includes the full transcript, a summary and a suggested next step so your team can pick up without asking the customer to repeat themselves.

Recovery loops

Instead of saying "I don't know" and leaving the customer stuck, the assistant tries harder. It rephrases its search, asks a clarifying question, offers a form to fill in or connects them with a human. If the same gap keeps coming up, it logs it so your content team can fix it.

Ready to build yours?

Start with this example and customise it to fit your organisation. You'll have a working assistant in minutes.