Pitch 1 of 3 · Bet hardest

Tracks.
The commentary track is back.

Creator commentary, friends' voice notes and short lessons, locked to the exact second of the show you're watching in its own app on your TV. You pause, it pauses. You rewind, it follows. No service has to build anything.

01

The demo we'd film

Netflix · Glass Onion · 47:12
Track · Rian Johnson commentary
"Rewind twenty seconds. Watch the glass on the left."
Rewind 20sKeep going
Scene Mail · from Brett
Left here Sunday, 0:14 · "I called it right here"

Mockup. Rian Johnson's Glass Onion commentary really exists. Netflix released it as a "Watching With…" podcast that you start on a countdown (JOE; The Verge).

  1. Friday night, Glass Onion on Netflix. One press and Rian Johnson's commentary plays from the TV, already in sync. No countdown.
  2. Brett pauses for a drink. Johnson stops mid-sentence. Play, and he picks up on the same word.
  3. Johnson says "rewind and catch this." On the podcast he takes that back right away because it would break the sync (The Verge). Here a card offers "Rewind 20s," the show jumps back, and the commentary stays locked.
  4. At 47:12 a second card appears: a voice note Brett left at that exact line last Sunday.
02

The problem

The commentary exists. It's just stuck.

Commentaries were recorded for "thousands of movies, television shows, and even video games," and most stayed on the discs when the industry moved to streaming. The Criterion Channel is the main exception (The Verge). Directors want them back: "Streamers have the ability effortlessly and for free to add [commentary]," Mike Flanagan told The Verge.

Even Netflix syncs by hand.

Netflix puts director commentary out as a podcast. You press play on a countdown, and every time you pause the movie you have to pause the podcast too. "Trying to sync a podcast in one app with a movie in another is not fun" (The Verge).

The best existing fix is a microphone.

RiffTrax Sync listens through your phone and needs 10–30 seconds after dialogue to lock on. Its own FAQ warns that streaming versions can break sync (RiffTrax).

Friends don't watch at the same time.

Disney+, Prime Video and Viki all shut down their synced watch parties, and Viki cited "decreased demand" (IGN, How-To Geek, Viki). The moment you want to share happens on different nights for different people.

03

The solution

A small app on the TV reads what's playing and the exact position, then plays a track pinned to that moment.

Format 1

Commentary

Audio from a creator, critic, rewatch-podcast host or the filmmakers, cut against a specific title and edition. It follows pause, seek and app switches.

Format 2

Scene Mail

GPT's idea. Leave a voice note or a reaction at a moment, and a friend gets it when they reach that moment, whatever night that is. No scheduling and no watch party.

Format 3

Lessons

GPT's beachhead. A film teacher's edition pauses your copy of the movie, explains a scene, poses a question, then resumes. Later: jumping between scenes across apps, once seeking proves reliable [unverified].

How it works

04

Why now

05

Who wants it

55%of Americans 12+ consumed a podcast in the past month, about 158 million people. Edison Infinite Dial 2025. source
$4.99what RiffTrax charges for one synced audio riff, a working price point for "audio you play over your own copy." source
Unmeasuredhow many people want synced commentary on the TV. No one has published it. Measuring it is what the beta is for.

The honest headwind: "If we look at Gen Z, this is an audience that is not necessarily growing up with director's commentary" (Julia Alexander, Parrot Analytics, in The Verge). The bet is that removing the sync hassle, plus notes from friends, widens the audience beyond film buffs.

06

How it makes money

Paid tracks

Sold one at a time, anchored on RiffTrax's $4.99. The creator keeps most of it. I'd propose a 70/30 split as a starting point, not a benchmark.

Creator memberships

Rewatch shows already sell bonus content. A watch-along track is a natural perk for paying members, and we take a platform fee.

Rights-holder licensing

Studios put vault commentaries and talent tracks on the layer, or a streamer licenses the sync instead of rebuilding its player. The exit path is a streamer, a podcast network or a studio's home-entertainment group.

07

Competition

ProductHow it syncsWhere it falls short
RiffTrax SyncPhone microphone10–30 s to lock; own comedy catalog only; phone only
CommontaryAudio fingerprint, user-made feedsSecond screen; adoption unclear [unverified]
Criterion ChannelCommentary built into its own playerIts own catalog only
Netflix "Watching With…"Podcast plus a countdownManual; breaks if you pause or rewind (Verge)
Prime Video X-RayFirst-party trivia and recapsPrime only; no third-party voices
Teleparty / RaveLive group syncTeleparty has no TV app; Rave plays inside its own app; neither does tracks

The open gap: nobody plays third-party tracks locked to the TV's native apps, on the TV itself.

08

The unfair advantage

The bottleneck is supply and rights, not code. That's your world.

Who owns a 2004 commentary? "There's like 80 different parties that own 80 different facets of this" (The Verge). Which rewatch hosts would cut a paid watch-along version? Which streamer's marketing team wants its back catalog rewatched with the talent in the room? Those questions get answered through relationships, and an engineer starting from zero doesn't have them. The code is a few weeks of work. The catalog is the moat.

09

Risks, and what you'd have to believe

You'd have to believeIf it's wrong
Netflix, Max and Prime publish an accurate playback position on Google TV that another app can read [unverified]The TV-native version is dead and Tracks shrinks to microphone sync, which is just RiffTrax. This is the kill test.
A track can play over another app without pausing it [unverified]Fall back to phone or earbuds. That works, but it's less magical.
Edition and ad-break differences can be handled with offsets [unverified]Start with ad-free tiers and single-edition titles
Reading now-playing data and using standard media controls isn't "automated means" under Netflix's terms [unverified]Get a legal read before launch. Never click inside the app.
Reach beyond Google TV (about 13% of US smart TVs, per Kagan) is possible: a phone companion, or Roku's local control protocol reporting position [unverified]It stays a Google TV niche until a platform partner shows up
Synced commentary is more than a cinephile nicheRiffTrax-sized: a nice business, not a big one
Original commentary is clear to sell (the RiffTrax model) and vault commentary can be clearedLaunch with original creator tracks only
10

The weekend proof

Saturday afternoon on your Sony. A sideloaded Google TV app reads Netflix's now-playing position and plays the Glass Onion "Watching With…" episode locked to it, for personal use only. Script: 30 minutes of the film, 5 pauses, 3 rewinds, and one trip to the home screen and back. Then repeat 5 minutes each on Max, Prime Video and YouTube to fill in a coverage grid.

Pass

The track re-locks within half a second after every pause, rewind and app switch, with no manual fixes. Brett asks to keep it on without being prompted. At least two of the four apps report position.

Fail → kill

Netflix and at least two of the other three don't expose a usable position, or drift is over 1 second after rewinds. In that case the TV-native thesis is dead, and we don't build a worse RiffTrax.

11

90 days

Days 1–14

Prove it

Weekend proof. Coverage grid across 6 apps. An offset tool for editions. A Scene Mail prototype between your TV and one friend's.

Days 15–60

Supply

3 creators, 5 tracks: one commentary, one rewatch-podcast edition, one film lesson. Scene Mail across 20 pairs of friends.

Days 61–90

Demand

Private beta with about 100 households from the creators' audiences. Measure the share who finish a track they start, reuse within two weeks, Scene Mails sent, and a price test at $4.99. Hold one studio library conversation. Go or no-go.

Go if most people who start a track finish it and creators ask to make a second one. Set the exact thresholds before the beta opens, so the result can't be argued after the fact.

The ask

One afternoon, three intros, then a decision.

  1. Decision: green-light the weekend proof on your TV and your Netflix. No money.
  2. Time: three intros. A rewatch-podcast host, a film critic or teacher with an audience, and someone who runs library or home entertainment at a studio. Thirty minutes each.
  3. Money, only if the proof passes: a contract Android TV developer for the 90 days. My rough estimate is $10–20k, to be scoped after the coverage grid. It's a guess, not a quote.
  4. The bigger call, at day 90: is this your company, a side bet, or something to hand to a founder you back?
Prepared for Brendan Brady · Oct 18, 2026 · Synthesis of four AI inventor outputs (Claude, Grok Bot, Gemini, GPT) plus independent checks. No invented metrics; anything not confirmed is marked [unverified]. Not legal advice.