5 Ways Slate's AI Features Are Built for Online Learning

AI content generation tools are everywhere. Most of them treat eLearning the same way they treat blog posts, marketing copy, or documentation: take a prompt, produce text. The output might be well-written, but it is not designed for learning.

The question for course creators is not whether AI can write. It is whether it understands how people learn. How content should be sequenced. When to introduce an interaction. How to write an assessment question that tests comprehension rather than recall. How to pace a 30-minute course differently from a 10-minute one.

Slate's AI features are purpose-built for these problems. Here are five ways that shows up in practice.

1. Instructional design methodology is built in, not bolted on

When Slate's AI generates a course, it does not simply write content and divide it into pages. Every generated course follows Gagné's Nine Events of Instruction, a pedagogical framework that defines how effective learning experiences are structured:

• Course openings gain attention with a provocative question or relatable scenario, state learning objectives explicitly, and connect the new topic to something the learner already knows.
• Each section presents content in atomic lessons (one core concept per lesson), provides guidance through pro-tips and callouts, and elicits performance through knowledge checks.
• Course conclusions provide feedback that explains why answers are correct, assess performance with a summative quiz, and enhance retention with key takeaway summaries.

This is not a surface-level label. The framework determines the actual block sequence, the types of interactions placed at specific intervals, and the structure of every section the AI produces.

Multimedia principles and interaction cadence

The writing itself follows Mayer's Multimedia Principles: second-person language to personalize content, text blocks capped at 150 words to support cognitive segmenting, bold key terms and headers for signalling, and active voice throughout. These are research-backed guidelines for reducing cognitive load in learning contexts.

Interaction cadence is enforced automatically. Every 3-4 text or image blocks must be followed by an interactive element: an accordion, a tabbed section, a knowledge check, a flip card, or a card carousel. Courses over 30 minutes include reflection prompts every 10 minutes. This prevents what instructional designers call "page-turner fatigue," where learners click through content passively without engaging.

Seat time is calibrated to reading speed standards for eLearning. Our seat time calculator uses 180 words per minute as the standard rate for general corporate training, with conservative (150 wpm) and quick (220 wpm) options depending on audience and content complexity. The AI uses these benchmarks to size courses appropriately: a 30-minute course targets 3-4 sections and 8-12 lessons, scaled to the selected duration.

Constructive alignment in assessments

AI-generated assessments follow constructive alignment principles. Every question maps directly to a stated learning objective. Correct feedback explains why the answer is right, connecting it back to the concept. Incorrect feedback guides the learner toward the right understanding without giving away the answer.

This is the difference between testing whether someone can recall a fact and supporting them in actually learning something.

2. AI that asks before it builds

The fastest way to generate a bad course is to describe a topic and hit generate. Slate's AI treats requirements gathering as a design step, not a hurdle to skip.

The conversational requirements phase

When you start a new course in Slate, the AI does not immediately produce content. It acts as an instructional designer in a kickoff meeting, asking 1-2 questions at a time about:

• The topic and subject matter
• Who the learners are and what they already know
• What learners should be able to do after completing the course
• Target duration and scope
• Whether to include assessments
• Preferred tone: formal, conversational, or approachable

The AI only marks the course as ready to generate when it has a clear topic, a defined target audience, 2-3 specific learning objectives, and a realistic scope. This mirrors what a human instructional designer does before creating anything.

Lesson-level drafting within course context

The same conversational approach applies to individual lessons. When drafting a new lesson within an existing course, the AI receives the parent course context (title, audience, objectives) so it maintains tone and complexity consistency. Lesson depth is configurable: overview (4-6 blocks), moderate (6-10 blocks), or comprehensive (10-14 blocks), letting you control granularity at the lesson level.

Why this matters

The gap between "generate a compliance course" and a well-scoped course with clear objectives, appropriate depth, and aligned assessments is enormous. The conversational approach closes that gap before any credits are spent. The entire requirements phase is free. You only commit credits when the requirements are right.

This is available on every plan, including Free. See pricing for credit allocations by tier.

3. Start from what you already have

Not every course starts from a blank page. Organizations have existing training materials: slide decks, policy documents, procedure manuals, onboarding guides. Slate's document attachment feature lets this existing work inform AI generation rather than starting from scratch.

Supported formats and how extraction works

You can attach PDF, PPTX, DOCX, TXT, and Markdown files (up to 20MB) as source material during the course or lesson drafting conversation. The AI extracts five categories from your document:

• Main topics: 5-15 key subjects identified in the material
• Learning objectives: 3-8 objectives derived from the content
• Core concepts: Essential ideas, definitions, frameworks, and terminology
• Structure: How the original content is organized
• Key details: Facts, figures, examples, and case studies worth preserving

This structured summary is injected into subsequent generation prompts, so the AI builds from your source material rather than its general knowledge.

From reference material to structured course

Presentations get dedicated handling. When the AI processes a slide deck, it reconstructs the logical flow and expands upon the talking points to create a comprehensive course. The slides provide the content backbone; the AI adds the learning structure: interactions, knowledge checks, proper pacing based on Gagné's framework, and the interaction cadence described above.

This also reduces the "hallucination" concern that comes with AI generation. When the AI has your source documents, it stays closer to your actual content rather than inventing details.

If you are coming from Articulate Rise, Slate also supports direct Rise course imports with structure, content, and media preserved.

4. Think first, create when ready

Slate's Claude connector (via MCP) enables a workflow that separates analysis from creation. This distinction matters more than most people realize.

The problem with jumping straight to generation

When someone has access to an AI course generator, the instinct is to describe a topic and start building. But the most impactful step in instructional design happens before any content is created: analyzing the learning need, understanding the audience, mapping competencies, identifying what the training actually needs to accomplish and why.

That analysis work is sophisticated. It involves reviewing existing materials, understanding organizational context, identifying gaps in current training, and determining what success looks like. Skipping it produces courses that are technically complete but strategically misaligned.

Claude as the analysis layer

With the MCP connector, Claude Desktop or Claude Code becomes a thinking partner for this upfront work. Users can have Claude analyze learning needs, break down competency frameworks, map curriculum sequences, and evaluate what a training program should cover.

Crucially, this analysis phase is not disconnected from the authoring tool. Users can talk to their existing Slate course library during this phase: "What compliance training do we already have?", "Show me courses tagged for onboarding," "What gaps exist in our Q1 curriculum?" This context makes the planning far more grounded, because the analysis accounts for what already exists rather than treating every course as a greenfield project.

When the analysis is done and the direction is clear, users invoke Slate's create_course or preview_course_outline tools to build from that foundation. The same conversation that identified the need can produce the course, without switching tools or re-entering context.

Course management through conversation

The MCP connector extends beyond creation. Users can list and filter courses, tag in bulk, create tracked sharing links, open review sessions, and check engagement analytics, all through natural language. The ability to query and manage an existing course library through conversation turns the connector into a content management interface.

For a deeper look at the MCP connector, see how to create eLearning with AI assistants. For how Slate's data model supports this kind of management at scale, see Slate as a content management system. Setup instructions are on the MCP page.

5. Repurpose existing content for a new medium

Here is a common scenario: an organization has a library of Canva presentations used for instructor-led training. The content has been reviewed, approved, and delivered successfully in person. Now they want measurable eLearning follow-up: a self-paced reinforcement activity that learners complete after the live session, with completion tracking and assessment scoring through their LMS.

Rebuilding that content from scratch in a different tool has historically been the only option. The Slate for Canva app changes that.

Canva presentations to SCORM courses

The Slate for Canva app works inside Canva's editor as a sidebar. Users select a presentation, configure course parameters (title, target duration, audience, learning objectives, tone, whether to include assessments), and the AI generates a full interactive course. Content extraction is intelligent: it reads slide content, reconstructs the logical flow, and adds learning structure using the same Gagné framework and interaction cadence as any other Slate-generated course.

The result is a complete, interactive course saved to your Slate library, ready to export as SCORM for your LMS.

Why this matters for L&D teams

Many organizations have extensive libraries of presentation-based training materials that have already been reviewed and approved. Converting these to measurable eLearning (with SCORM tracking, completion status, and assessment scoring) has historically required rebuilding from scratch in an authoring tool.

This integration creates a bridge between presentation-based ILT content and self-paced digital learning. The content is already approved. The learning objectives are established. What changes is the medium and the measurability: learners complete an interactive follow-up activity, and the LMS records their progress and scores.

The reverse flow also exists. Through the MCP connector, users can export Slate course content back to presentation format for instructor-led sessions, creating a bidirectional pipeline between in-person and digital delivery.

What this adds up to

These five features share a design philosophy. AI in eLearning should understand how people learn, not just how to write. The pedagogical frameworks, the conversational requirements gathering, the document-informed generation, the analysis-before-creation workflow, and the content repurposing pipeline all serve the same goal: enabling course creators to produce learning experiences that are pedagogically sound, not just quickly produced.

Speed matters. So does quality. The point is that you should not have to choose.

Try it

Sign up free to start building courses with the conversational AI workflow and document uploads included on every plan. See pricing for credit allocations and plan details. Connect your AI assistant via the MCP connector on Standard and Pro plans. And keep an eye on Slate for Canva for the presentation-to-course pipeline.