New section type
Build software walkthroughs that survive the next UI update.
Screen Demo turns screenshots into interactive software training - with four learning modes, AI-written instructions, and no re-recording when the interface changes.
Recording-based tools have one problem.
Every time the UI changes, the recording is invalid. Someone re-records, re-cuts, re-encodes - then re-records again for every language you support.
Screen Demo's source of truth is a set of static screenshots and hotspots. Swap one image, nudge the hotspot. Everything else - instructions, narration, annotations, translations - stays intact.
One build. Four learning experiences.
The author builds once. The learner picks the mode, or the author locks a specific mode for compliance-critical content. No re-authoring per audience.
Watch it
For: Onboarding and awareness
An animated walkthrough: cursor glides step to step, callout explains each screen, optional narration reads it aloud. Zero effort from the learner.
Show me
For: Guided practice
The hotspot is visible. The learner clicks through at their own pace - no automation, no assessment pressure. The right step for a first attempt.
Try it
For: Assessment and compliance
Hotspot hidden. A task banner tells them what to do. Misses are tracked. A hint can reveal the hotspot after a set number of wrong clicks.
Watch then try
For: The instructional design default
The learner watches first, then immediately performs the same task. The sequence the research recommends - and it's one option, not a separate build.
Why it's different.
Screenshots, not recordings
Swap one PNG and adjust a hotspot when the UI changes. No re-recording, no re-cutting, no re-encoding. Everything attached to that step - narration, annotations, translations - stays intact.
AI drafts the writing
Once three steps are placed, per-step instructions and Try it prompts unlock. CourseAgent reads the actual pixels and writes everything in the course's tone, audience level and reading age. You edit; you don't start from blank.
One build, four modes
Watch it, Show me, Try it and Watch then try all come from the same steps. No forking, no re-authoring per audience or per mode.
Everything translates
Instructions, callouts, banners and narration scripts translate with the rest of the course into any of the 19 supported languages - including localised narration voices. No re-recording per language.
How to build one.
Three steps. The AI handles the writing once you've placed a few hotspots.
1. Capture
Screenshot the UI using the built-in browser capture - no extension or desktop app needed, just the native browser screen-sharing API. Or paste from clipboard (Cmd/Ctrl+V), or upload a PNG or JPG from another team. Auto-cropped to 16:9 in the same modal.
- Native browser capture
Pick source once, keep capturing frame after frame without re-picking.
- Paste from clipboard
Cmd/Ctrl+V from macOS Cmd+Shift+4 or Windows Snipping Tool.
- Upload PNG or JPG
Drag and drop pre-recorded product screens from another team.
2. Place the hotspot
Draw a rectangle over the click target. Coordinates are stored as percentages of the image - the hotspot survives every later resize or theme swap without drifting.
3. Let the AI write it
With three or more steps in place, draft per-step instructions, Try it prompts and narration in one click - grounded on the actual pixels, in the course's tone. Draft the intro and completion screens whenever you're ready.
Who it's for.
Learning teams
One build produces four learning experiences - no forking, no re-authoring per audience. AI drafts the writing grounded in actual pixels and the course's tone. Everything translates with the course, including narration, into all 19 supported languages.
Enablement, product marketing and support
Product screens change every sprint. Update the walkthrough in one place - non-technical authors maintain it without a video editor, voice-over artist or design re-mock.
Compliance and audit
Try it with no-hints completion evidences that the learner performed the task independently, with a recorded miss count. Every step is a static, auditable asset - not an opaque video file. Pass rules plus the standard SCORM completion pipeline give a defensible attestation of task completion.
When Screen Demo wins - and when it doesn't.
The honest version. Recorders are faster on one narrow shape of content. Everything else, the screenshot-first flow beats them - especially on the second edit.
Screen Demo is the faster choice when
- The procedure is 6-20 steps and will need editing after first publish.
- The UI is behind SSO, a kiosk, a hardware panel or a third-party portal you can't screen-record cleanly.
- The content will be translated - AI text and narration in 19 languages vs re-recording per language.
- Any step will change. Swap one image, keep every annotation, narration and translation intact.
A screen recorder may be faster when
- The first build is a long linear procedure (30+ steps of pure click-through with no explanation between them), and you're confident it won't change. A single recording pass beats 30 individual captures - once.
How Screen Demo compares.
Focused on the software-simulation use case.
| Feature | CourseAgent | dominKnow ONE | Adobe Captivate | Storyline 360 |
|---|---|---|---|---|
| Screenshot-first (no recording required) | recording-first | recording-first | recording-first | |
| Modes from a single build | 4 | 5 (recording-first) | 3 - Demo, Training, Assessment | via separate slides |
| AI-drafted step instructions from screenshots | vision-grounded | text assist only | ||
| AI translation incl. narration voices (19 languages) | text only | manual per language | manual per language | |
| Author skill required | low, non-technical | medium | high | high |
| Editable after publish without re-recording | swap PNG, move hotspot | re-record required | ||
| Runs inside an all-in-one authoring + LMS platform | authoring only | standalone | standalone |
Yes Partial No
Start your 14-day free trial today.
Full Professional access. No credit card. No technical skills required.
Just describe what you want to teach.
See what's included in the 14-day trial →
CourseAgent is in beta - read this first or join the launch waitlist.