SaaSVideo EditingAI Transcript

Capti: Final Cut Pro Captioning Tool

A complete short video captioning platform for Final Cut Pro users designed for content creators uploading on Social Media Platforms.

Visit project
Capti: Final Cut Pro Captioning Tool — 1
Capti: Final Cut Pro Captioning Tool — 2
Capti: Final Cut Pro Captioning Tool — 3
Capti: Final Cut Pro Captioning Tool — 4

TLDR;

Capti is a captioning tool for Final Cut Pro that does what Apple won't: multilingual transcription, customizable titles (not boring subtitles), and intelligent word grouping that actually looks like what you see on YouTube or TikTok.

Upload a video, get styled captions back as FCPXML. Drag into your timeline. Done.

With a free tier and affordable subscription, it's the tool FCP editors didn't know they were missing.

The issue

Final Cut Pro added automatic transcription in a recent update, but it's English only and produces flat, uncustomizable captions: not the styled titles creators actually want.

So editors are stuck with bad options:

  1. The CapCut detour: edit in FCP, export, import into CapCut, generate captions there (paid), export again. A workflow that makes you question your life choices
  2. Legacy Mac apps: expensive subscriptions for software that looks like it was designed in 2014 and barely works
  3. Manual captioning: placing titles word by word. Nobody has time for this

Capti is the modern, affordable alternative: fast transcription, smart grouping, real customization, and a direct FCPXML export that drops right into your timeline.

Process

It started as a Python script. One evening, a proof of concept: upload a video, get a transcription, export it. The results were good enough to prove the idea had legs.

Key milestones from prototype to product:

  1. Local to cloud: moving from a script on my machine to a full web platform with auth, storage, and billing without losing in performance
  2. Transcription provider odyssey: tested many providers, landed on AssemblyAI after OpenAI's Whisper stopped getting updates and newer models dropped timestamp support
  3. Smart grouping: automatic word grouping with AI inference, switched from OpenAI to Groq for speed
  4. Web video timeline: built a fully interactive timeline in the browser for previewing and adjusting captions before export
  5. FCPXML export: the hard part. Ensuring timing accuracy, timeline superposition, and title attribute consistency so the output just works in Final Cut (this took a lot of .fcpxml reverse engineering due to the lack of documentation)
  6. Billing without the tax nightmare: implemented Polar.sh as Merchant of Record to avoid declaring VAT in every customer's country
  7. Documentation: comprehensive getting started guide and troubleshooting

Built during evenings from August to December 2025.

The stack

Capti is built on a modern web stack optimized for real-time processing:

  • Next.js: React framework with App Router for the platform UI
  • Convex: Real-time backend: database, reactive queries, and mutations
  • Clerk: Authentication provider
  • Shadcn/ui + TailwindCSS: Design system with accessible components
  • Polar.sh: Merchant of Record (Stripe wrapper handling global tax compliance)
  • AssemblyAI: Multilingual AI transcription with word-level timestamps
  • Loops.so: Transactional emails and newsletter delivery
  • Vercel: Hosting and deployment
  • Vercel Blob: Object storage for video uploads
  • PostHog: Analytics and session replays
  • Fumadocs: Auto-generated documentation from Markdown files

Communication and Marketing

FCP editors live on Reddit and Facebook groups. That's where they ask questions, share workflows, and complain about captioning.

Discovery happens through:

  • f5bot alerts: a Reddit bot sends me an email whenever someone posts about captioning in FCP communities. A helpful, non-spammy comment is usually enough to spark interest
  • Word of mouth: editors share tools that save them time
  • Documentation as marketing: the getting started guide doubles as a landing page for search traffic

Once a user signs up, retention is everything. Gathering feedback early, fixing pain points fast, and making people feel heard is what keeps a small product alive.

Deliverables

  • The Platform: fully functional SaaS with 30 users, transcription pipeline, timeline editor, and FCPXML export
  • Documentation: comprehensive guides and troubleshooting powered by Fumadocs
  • Brand identity: logo, visual language, and consistent design system
  • Community presence: active Reddit engagement and user feedback loops