Programmatic (and self-updating) SaaS demo videos
Automated Playwright-based demo recording with AI voiceover; solves real help-center churn.

Playwright automation for demo recording, but Loom and Arcade already capture this with less technical setup.
SaaS product teams, help-center creators, technical documentation owners
Loom · Arcade · Descript
PS: Because this is heavy on infra we have a waitlist thing but I'm monitoring and will let you in within minutes/hours, just need to space people out a bit. Thanks
Automated Playwright-based demo recording with AI voiceover; solves real help-center churn.
Standardizes portable cryptographic receipts for agent behavior—but adoption unclear, overlaps Nobulex heavily.
SkillForge turns the old 'show, don't tell' trick into code: record a task, and their AI teases clicks, keystrokes and navigation out of pixels into a stepwise skill file you can edit and export. The ability to trim video, rewrite steps via AI, and output a SKILL.md for agent frameworks is a practical, opinionated workflow that could shortcut lots of brittle RPA scripting — my main questions are reliability across dynamic UIs and privacy/recording controls, but the product direction is smart and tangible.
It records you doing a task, then auto-extracts clicks, keystrokes and navigation into a human-editable, step-by-step workflow and a SKILL.md you can feed to agent frameworks — that demo-to-skill UX is a real 'oh nice' moment. The landing page shows practical examples (spreadsheet entry, research, crypto checks) and an inline editor, but I want clarity on robustness: how it handles dynamic selectors, cross-app gestures, and sensitive data in recordings.
Records clicks, keystrokes and app switches in-browser, then extracts them into an editable, step-by-step workflow you can trim, reorder and export as SKILL.md for agent frameworks. The UX reads thoughtfully — no-install browser recording, review/edit and Markdown export are practical — but the page shows polished examples rather than edge-case behavior; robustness across UI changes, element matching and secure credential handling are the unanswered hard problems here.
Lets agents actually see the screen and act on it by returning OCR text with pixel coordinates and offering commands like click_at, type_text, and press_key. You can run it instantly with npx (it auto-creates a Python venv and hooks into Apple Vision/Quartz), and there are ready-made integration snippets for Claude, VS Code, and Cursor — a pragmatic, technically neat tool for closed-loop agent UI work. It’s limited to macOS 13+ and Apple APIs, but within that niche it removes a lot of friction.