Turn live websites into editable demo sequences, then export polished MP4s. Live capture, heuristics templates, visual scenes and ffmpeg export, all on your desktop.
Seqyra records your app: screenshots, clips, whole flows. But what you end up holding isn't a flat take. It's a sequence: each capture, caption, zoom and pause is a step with its own timing you can reorder, re-time or re-run on its own.
So fixing one moment means editing one step, not starting the recording over.
One URL in, one finished MP4 out: capture, steps, templates, visual scenes, timing and export, with nothing hidden behind a cut.
Capture, sequence, polish, export. Each stage hands off to the next in one deterministic pipeline, with no cloud round-trip and no black boxes.
Live viewport preview in a real browser. Grab manual or interval PNG screenshots, or hybrid video clips for clicks, typing and scrolls.
Ordered actions: goto, click, input, scroll, wait, screenshot, caption and zoom, targeting the viewport or an element via CSS, text, XPath or aria. Each step gets a thumbnail after you run it.
The heuristics engine inspects the live page with rules rather than a model: it classifies the site, scores CTAs and forms, dismisses cookie banners, then builds tailored step sequences you can load and edit.
On screenshot and caption steps, open the Visual panel: fixed 1440×900 scenes with canvas backgrounds, text layers, shapes and drawing tools, rendered the same in editor, preview and export.
Per-segment duration and transitions on the timeline. Add music and sound FX on dedicated lanes, trim and fade clips, and keep the preview clock aligned with ffmpeg export.
Export a polished MP4 with ffmpeg, locally on your machine. Video segments, visual scenes, music and SFX are mixed into one deterministic export.
Paste a URL and the engine waits for network and layout stability, reads DOM signals (CTAs, nav, forms, headings), classifies the site type, then generates up to four editable variants. Rules, not a model, so it needs no AI setup and makes no external calls. You can still switch AI on separately when you want a faster start.
Primary flow, fast pacing, minimal waits.
Same flow with added waits and zoom-in steps between interactions.
A second flow when the page has more than one obvious demo story.
Screenshots and navigations only, capped at six steps.
Learn the UI once and reopen projects any time. Optional AI is there when you want to move faster, with guides for a local model or your own agent.
Runs on your machine. No account required. Everything stays local. Only downloads are tracked.
Having trouble? View the installation guide
Loading latest version…