Capturing a screen and a webcam in a browser and exporting an MP4 is the quick part. What makes Tella feel like Tella is the editing layer: clips on a timeline, cuts made by deleting words in the transcript, silence and filler-word removal, layouts, backgrounds and zooms. An agent can build a rough version over a weekend from MediaRecorder, whisper.cpp and ffmpeg. The 4K and 60 FPS rendering, the polish and hosted share pages with viewer analytics take longer and need a server and bandwidth, and open-source recorders already cover much of this.
Before you buildTella has no free plan, but Cap, OpenScreen and Screenity are free recorders with polish settings. If you only record occasionally, try one before building.
Build me a local screen-recording editor to replace Tella for short product demos. Requirements: - Node.js 22 + TypeScript: a Fastify server and a React UI built with Vite. `npm start` serves http://localhost:4740; use a Chromium-based browser. - Recorder page: getDisplayMedia for the screen, getUserMedia for microphone and camera, a 3-second countdown, pause and resume. Each take is a clip, saved as webm chunks under ./data/clips/<id>/ while recording. - Storage: better-sqlite3 in ./data/studio.db: clips, word-level transcripts, edits and exports. An edit is a JSON list of kept time ranges, zooms and a layout per clip, saved after every change. - Transcript: run whisper.cpp (WHISPER_BIN and WHISPER_MODEL in .env) with word timestamps. Deleting words in the transcript view removes those ranges from the edit, with undo and redo. - Cleanup buttons: cut silences longer than 0.7 seconds and cut filler words from a list I can edit (um, uh, you know). - Layouts: screen only, screen with a round camera bubble in a corner, and screen beside camera, over a background colour or gradient with padding and rounded corners. Add zoom regions by dragging a box on the timeline. - Export: ffmpeg renders the edit to H.264/AAC MP4 at 1080p and 30 FPS with a progress bar, and optional burned-in subtitles from the transcript. Serve /v/<id> with a player and the transcript; ids are random 12-character strings (unlisted). - Single user. No accounts, no telemetry, local-first: recordings never leave this disk. - Out of scope: 4K or 60 FPS, auto-zoom, a native recorder, an AI chat editor, comments, viewer analytics, custom domains, teams. - README: installing ffmpeg and whisper.cpp, expected render time for 10 minutes of video, sharing through a tunnel, and a warning that a browser cannot see the cursor outside the shared tab, so zooms are manual.
Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.
The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Tella.
Add video import to the screen-recording editor in this folder. Requirements: - An Import button and a command `studio import <file...>` that accept .mp4, .mov and .webm. - Copy each file to ./data/clips/<id>/source and use ffprobe to read duration, size and whether there is an audio track; reject files without video. - Queue transcription for each import and show progress on the clip list. - Keep the original file untouched and name the clip after the filename. - Update the README with the supported formats and the Tella batch export tip.
Add click-based zooms to the screen-recording editor in this folder. Requirements: - Add a record mode "this tab" using getDisplayMedia with the current tab preferred; in that mode log pointer position and click times inside the recorded page with performance.now() offsets. - Save the log next to the clip. In the editor, a button "suggest zooms" creates a zoom region (1.6x, eased in and out over 0.4 seconds) for each click that is more than 2 seconds after the last. - I can accept, move or delete each suggestion; nothing is applied without my accepting it. - Cover the suggestion logic with tests, including clicks at screen edges. - Update the README with why this works only for the recorded tab.
Add titles and chapters to the screen-recording editor in this folder. Requirements: - After transcription, send the transcript to an OpenAI-compatible endpoint from .env and ask for a title under 60 characters and 3 to 8 chapters with start times in seconds. - Show them as editable fields on the clip; chapters appear on the share page as clickable timestamps that seek the player. - Validate the JSON reply and clamp chapter times to the video length. - Keep it optional: without the endpoint set, show empty fields. - Update the README with what text leaves the machine.
People pay for videos that look finished without learning an editor: layouts, zooms and backgrounds come as presets, and a share page with analytics sits on top. Reddit comparisons call it the all-in-one pick with nicer visual options than Loom, while one commenter found the open-source Cap unstable.
Estimate · Recording, transcribing and rendering all run on your machine at no cost. Sharing videos needs a tunnel or a small VPS, and storage and bandwidth grow with what you record. The build is token usage on your agent plan.
| Plan | Monthly | Annual, per month | What you get |
|---|---|---|---|
| Pro | $26 | $13 | Unlimited videos, clips up to 6 hours in up to 4K, 4K export, 106 transcription languages, 3,000 Tella Agent credits per user a month, team collaboration, analytics history of up to 30 days. |
| Premium | $39 | $19 | Everything in Pro plus 6,000 credits per user a month, custom domain, domain restriction, Tella branding removed, 4K sharing, 60 FPS export and all-time analytics. |
| Enterprise | custom | — | Everything in Premium plus SSO and SCIM, a shared Slack channel, onboarding, volume discounts and a SOC 2 Type II report. |
No free plan: new workspaces get a 7-day trial of Pro that exports HD at 30 FPS, caps clips at 2 hours and includes no AI credits.
Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: The Tella Agent, an AI editor that makes edits from a chat, Native macOS and Windows recorders and the Chrome extension.
About $26 a month on the Pro plan per seat, $312 a year (checked 2026-10-19).
The Tella Agent, an AI editor that makes edits from a chat; Native macOS and Windows recorders and the Chrome extension; Custom domains, white-label pages, password protection and viewer analytics; Slack, Notion and Linear integrations, plus shared and team playlists; Auto layouts, auto zoom and styled subtitles. If any of those matter to you, keep paying.
Yes: Cap (open source), OpenScreen (open source), Recordly (open source).