What shipped in LearnBuilder this week: a calmer editor, smarter AI course creation, and 60 languages

This update is bigger than the usual weekly patch. A round of editor cleanup, several new layout primitives, two upgrades to the interactive slideshow builder, a meaningful expansion of how AI course creation handles instructional design and audience, full localisation across sixty languages, AI-translated video subtitles, and finer control over AI video generation. Most of it has been on the roadmap for a while; this is the week it landed together.

Below is what changed and, more importantly, what each change actually unlocks for the people using the tool day to day.

A calmer editor

The editor has been cleaned up. Toolbars that were always visible are now visible when you need them. Controls that overlapped with the content area have been moved out of the way. The default state when you open a lesson is closer to the published view, with the chrome receding until you start editing.

The benefit is less obvious in screenshots than in practice. For people who spend hours a day in the editor, removing visual noise is the difference between an interface that feels heavy and one that gets out of the way. It also makes review sessions easier — when you walk a stakeholder through a draft, they see the lesson, not the editing interface.

This is the kind of change that does not have a feature name. It is also one of the most-requested changes from the early users, so it is worth calling out explicitly.

Block backgrounds, full-width layouts, and proper spacing controls

Blocks now support backgrounds — solid colours, images, or gradients — and a full set of margin and padding controls. They can also stretch to 100% width of the lesson, breaking out of the central content column when the design calls for it.

In aggregate, this is the difference between content that looks like a default theme and content that looks like yours. A full-width image block with overlay text. A section break with a coloured background that signals a change of pace. A scenario set against a gradient that matches the topic. A breathing-room margin around a quiz so it does not feel jammed against the explanatory text above.

Brand colours and fonts are still configured at the account level and inherited automatically. The new controls give designers control over the per-block visual rhythm of a lesson without breaking the system.

Full-screen mode in the interactive slideshow builder

The interactive slideshow builder now has a full-screen editing mode. The rest of the application chrome collapses, the canvas expands, and you get the screen space back that complex interactions actually need.

This matters more than it sounds. Layered interactions — a branching scenario with multiple pathways, a simulation with three or four interactive elements, a custom layout with a dozen tracked components — were genuinely cramped in the standard editor. Full-screen mode gives you the room to lay them out without zooming in and out, and it keeps element selection and the property panel where you expect them. For anyone building anything beyond a one-element interaction, this is a quality-of-life change with real consequences for how much you can fit on a slide and still keep editing.

An SVG element with editable source code

Interactive slides now include an SVG element. Drop one in, paste the SVG markup, and the source is editable directly inside the element panel.

The reason this is in the product: AI tools and designer toolchains both produce SVG natively. An icon library, a custom diagram, an org chart, a process flow, a visual you generated with an external tool — all of these arrive as SVG. Until now, getting them into a lesson meant exporting to PNG, losing scalability, and editing the source elsewhere. Now you can paste it in and tweak it in place, change a colour, add an animation, fix a label. For technical designers, this is also the cleanest way to drop in a hand-built SVG that is reused across lessons.

This pairs with the existing AI coding block. SVG covers the static-but-editable visual case; the AI coding block covers the dynamic-interaction case.

Four instructional-design approaches for AI course creation

AI course creation now lets you choose between four instructional-design approaches when generating a course outline:

• Standard — a balanced structure with explanatory content, knowledge checks, and scenario activities sequenced toward the learning objectives.
• Action-first — opens each lesson with the task or behaviour the learner needs to perform, then fills in the knowledge required to do it. Good for skills training and procedural content.
• Objective-first — leads with the learning objective and structures content around evidence that the objective has been met. Good for compliance, certification, and assessment-driven programmes.
• Story-driven — frames the content around a narrative or scenario that unfolds across the lesson. Good for soft-skills training, ethics, and topics where context matters more than information transfer.

The benefit is that the AI is no longer making one set of structural assumptions on your behalf. The same source material — a policy document, a product spec, a subject-matter brief — produces meaningfully different courses depending on the approach. Designers who know which approach fits the audience can pick it directly, instead of regenerating with prompt tweaks until the structure feels right.

Experience level and learning context

In the same step where you choose the instructional approach, you can now set the experience level of the target audience (novice, intermediate, advanced) and the learning context in which the course will be used (onboarding, compliance, performance support, deep-skill development, refresher, and others).

This shifts what the AI generates in concrete ways. A novice-onboarding course gets more scaffolding, more retrieval practice, and shorter cognitive-load chunks. An advanced performance-support course skips foundational explanations and goes straight to decision support and edge cases. A compliance refresher generates differently from an initial compliance course, even on the same regulation.

The combination of approach + experience + context is what an instructional designer is doing in their head when they brief a developer. Surfacing it as explicit input means the first draft is closer to what the designer would have asked for, with fewer rounds of regeneration.

Sixty languages, with course translation

Course language selection now covers sixty languages, and full course translation between any of them is available with one action. Generated content respects the conventions of the target language — not just word-for-word translation, but appropriate idioms, examples, and instructional patterns.

For organisations with regional teams, this collapses what used to be a multi-week localisation pipeline into something a single designer can run from the editor. The translated course retains its structure, its branding, its interactive elements, and its accessibility metadata. You still want a human review pass for any high-stakes content, especially compliance, but the starting point is far closer to publish-ready than a machine translation of HTML output would be.

AI-translated video subtitles

Video subtitles can now be AI-translated alongside the course translation. Upload a video once with subtitles in the source language, and translated versions are produced for every other language the course is published in.

This is the second piece of the localisation story. Translated text without translated subtitles produces a course that is bilingual in the worst way — the lesson is in Spanish, the video is captioned in English. The subtitle translation closes that gap. As with course translation, a human pass is recommended for video where the script carries weight.

Start and end frame control for AI video generation

AI video generation now accepts a start frame and an end frame, in addition to the prompt. You upload an image of where the video should begin, an image of where it should end, and the model generates the motion between them.

The benefit is control over output that previously required prompt-and-pray iteration. For product walkthroughs, scenario reconstructions, or any video where the visual continuity with the rest of the lesson matters, you can now anchor the result to specific frames you have control over. The animation can pick up from a still you placed earlier in the lesson and end on a still you use afterwards. It is not the same as a hand-edited cut, but for the use cases AI video is realistically good for, this closes most of the gap between "approximately what I wanted" and "what I wanted."

If you come across anything in this release that does not behave the way you expect or have ideas for new features, contact us.

The new features are live for all accounts. If you want to try them on a real course, the free trial is the place to start.