If you’re shopping for a screen recorder in 2026, the market has converged on the same set of features: clip, share, AI transcript, AI summary, AI chapters. After testing Loom, Cap, Clipy, Tella, and Vidyard side by side, we noticed three features that were missing from every competitor. So we built them. They ship today in Pullsy Record — free for everyone.
1. Voice Clone Playback — listen to the recording instead of watching it
Open any Pullsy recording and click Listen (AI voice) next to the transcript. The video continues, but now a cloned voice reads the transcript in sync. The TTS spans the same time window as the speaker did — fast-forward through a 30-minute meeting, scrub to a specific timestamp, and the spoken audio stays locked to the video.
How it works under the hood: we send each transcript segment to OpenAI’s TTS-1 endpoint, render it as MP3, cache it on the Cloudflare edge, and pipe the bytes to the viewer’s <audio> element with currentTime sync.
Right now you can pick from 6 stock voices (alloy, echo, fable, onyx, nova, shimmer). Soon: upload a 30-second sample and we’ll match your tone, pace, and pitch.
<!-- The recording page has a single toggle -->
<button id="listenToggle">▶ Listen (AI voice)</button>
<audio src="/api/voice/segment/<id>/0?voice=nova" autoplay></audio>Why this matters:
- Long meetings → 1.5x speed audio without losing the meeting context
- Slack/iMessage previews → speaker attached, not just video
- Accessibility → “Listen to this recording” instead of “Watch it”
2. Auto-clip Highlights — AI picks the 3 best moments, plays them in sequence
Most recordings have 5 minutes of context the viewer doesn’t need. Wouldn’t it be nice if an AI made a 90-second trailer of just the interesting parts?
Click ✨ Make trailer on any Pullsy recording. Our GPT-4o-mini pipeline reads the full transcript and:
- Identifies the most surprising / aha moment
- Finds the concrete step / how-to part (actionable)
- Locates the conclusion / wrap-up
Returns 2-3 ranges of ~30 seconds each, with a 2-sentence hook narrative. The viewer has a ▶ Play trailer button that auto-advances through the moments. Each one is standalone — make sense without prior context.
Example output for an onboarding walkthrough:
AI Trailer · 70s
Discover the key steps to a successful onboarding process and learn about a surprising mistake that many teams make. This concise guide will help you streamline your setup and ensure a smooth transition.
▶ Play trailer
- Surprising Insight · 19s – 29s · 10s · This clip highlights a common oversight that can significantly impact the onboarding process.
- Implementing SSO · 5s – 12s · 7s · A straightforward explanation of the SSO setup process with practical steps to get started.
- Bulk Invites Setup · 50s – 56s · 6s · A practical walkthrough of the workspace invitation process.
(Yes — for an actual onboarding video, the AI picked the exact “you forgot to enable Just-In-Time provisioning” moment as the surprise. That’s the clip your CEO needs to see.)
3. MCP Server — your recordings are now AI-agent-accessible
If you use Claude, ChatGPT, Cursor, or any MCP-aware AI tool, you can now connect them directly to your Pullsy library.
The server lives at https://record.pullsy.online/mcp and exposes four tools over JSON-RPC 2.0:
| Tool | What it does |
|---|---|
record.list | List recent recordings with metadata. |
record.get | Get one recording with full transcript + AI features. |
record.search | Search across transcript text. |
record.transcribe | Run server-side Whisper on a recording. |
Example — ask your AI agent: “Find all my Pullsy recordings that mention pricing and give me a 1-paragraph summary of each.”
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "record.search",
"arguments": { "query": "pricing" }
}
}Pullsy is one of the first screen recorders on the public internet to ship an MCP server. Clipy has one; nobody else in the Loom / Cap / Tella / Vidyard space has shipped this yet.
Why these three?
We picked them because they’re sticky features that justify a paid tier without taking away anything from the free plan. The free plan stays free forever. The Pro tier gets:
- Voice clone from a 30-second sample (not just stock voices)
- Higher-quality trailer generation (60s+ videos, multiple language summaries)
- MCP server auth tokens + per-user rate limits
- Bulk operations (render voice for 50 recordings at once)
We’re not announcing pricing here. We’re announcing that Loom refugees have somewhere to go that doesn’t lock AI features behind a $15/user/mo paywall.
Try it now
- Open https://pullsy.online/record and record something.
- When it saves, open the share URL.
- Click ✨ Make trailer + 🎙️ Voice clone playback + Listen (AI voice).
If you’re building an AI tool and want to integrate the MCP server, drop us a line. The endpoint is open, JSON-RPC 2.0, no auth yet — please don’t abuse it.