Overview
Multi made Mac applications collaborative. A teammate could point, draw, or take control inside shared code, designs, and documents. The startup began as Remotion, a virtual office, then narrowed its focus to technical teams. On June 24, 2024, it announced that its team was joining OpenAI. Its app was scheduled to close one month later.[1]
Its strongest lesson concerns the difference between creating a habit and improving one. Remotion tried to generate spontaneous conversation. Multi improved the pairing calls engineers already held. That shift produced more concentrated use, but the public record does not establish a durable standalone business. The acquisition preserved a team’s opportunity to build while ending customers’ access to its product.
Image 1 / 1
Founding Story
Alexander Embiricos and Charley Ho met at Stanford. Embiricos worked in product at Dropbox. Ho worked at Bebop, then Google after its acquisition. Their investor Sarah Guo described the initial product as an always-present desktop dock showing who was available to talk. Guo led Greylock’s Series A and joined the board; First Round also backed the company.[2]
Development began in May 2019, according to engineer Dan Wood’s later account. Remotion initially used Apple’s established desktop frameworks. Its native Mac implementation was a product decision: an application meant to remain open throughout the day needed to feel at home on the desktop.[3][4]
The pandemic made the original problem visible, but enthusiasm inside the team became an unreliable guide. In the Engineering Founders episode notes, Ho describes the temptation of “just one more bug fix.” Embiricos states: “One more iteration isn't gonna help it.” These short excerpts capture the founders’ recognition that improving the existing product would not resolve its positioning problem.[5]
Timeline
- May 2019: The team starts writing the Mac application.[3]
- October 2020: Remotion launches its beta and announces $13 million raised across seed and Series A funding.[6]
- July 2021: Remotion 2.0 reaches first place on Product Hunt for the day, according to the company’s account.[7]
- January 2023: The changelog records Zoom infrastructure, a shared editor, and features for opening shared content locally.[8]
- July–August 2023: Remotion becomes Multi and launches its technical-collaboration positioning.[9][10]
- March 2024: Multi documents lower-latency screen sharing and its first AI summary features.[11][12]
- June–July 2024: The team announces its move to OpenAI and sets July 24 as the app’s sunset and data-deletion date, with extensions available by request.[1]
What They Built
The original Remotion placed colleagues on the desktop. Status indicators and quick calls aimed to lower the effort needed to start a conversation. Multi changed the unit of value from ambient presence to a useful working session.[2][9]
A participant shared an application. Others could point, draw, or interact with it. Shared windows behaved more like local applications than a presentation inside a meeting. Automatic links let viewers open the underlying files or documents. Shared notes helped participants retain the discussion. The August 2023 launch supported two simultaneous shares; later marketing advertised up to ten.[10][13]
This demanded more than a video-call wrapper. The remote-control engine distinguished local input from remote events so the machine’s owner could regain control. Multi used system accessibility permissions to deliver keyboard and pointer events.[14]
Legible text and fast response also required different video settings from ordinary conferencing. Engineers moved video toward peer-to-peer transport, adjusted encoding, and reduced buffering. Their March 2024 account reported shared-control streams 2.3 times faster. This was a company-reported engineering result, not an independent benchmark.[11]
AI notes extended the session into follow-up work. Multi put summaries inside a shared editor, where people could correct them. Its design account also separated shipped behavior from experiments, including proposed citations and contextual follow-ups.[12]
Market Position
Target Customers
The pivot targeted technical teams, particularly people pairing, reviewing code, or solving problems together. It also removed remote-only positioning: colocated engineers could benefit from working across separate screens.[9]
Market Size
The sources reviewed do not establish Multi’s addressable market or revenue. Remote-work growth explains interest in the category; it does not show how many teams would pay for a separate collaboration layer.
Competition
Slack, Teams, and Zoom already held everyday team communication. The founders’ account describes how their expanding audio and video features narrowed the space available to virtual offices. A new tool needed to improve a specific task enough to justify adoption.[15]
That distinction still matters. As checked in October 2026, Tuple sells dedicated pairing software with remote control, while Apple supports remote control during one-to-one FaceTime calls. Multi’s closure therefore leaves customers alternatives; it does not prove an empty market.[16][17]
Business Model
Multi’s surviving pricing page lists a free plan and a proposed Team plan at $30 per seat monthly. It also says free users receive Team features until pricing rolls out in 2024. Several paid-plan features are marked as forthcoming. This page establishes pricing intent, not collected revenue or paid conversion.[18]
No revenue, gross margin, burn, or acquisition price was established in the reviewed first-party record. Without those figures, funding cannot support a claim about investor returns or the economics of the exit.
Traction
Remotion’s launch retrospective reports 400–500 new teams from its Remotion 2.0 campaign, versus 141 from a later Spotify-sharing feature launch. The latter received less promotion. These figures measure acquisition events, not retained teams.[7]
A founder update reproduced on Ho’s profile reports weekly active users falling 8% after the January pivot, while screen-sharing usage rose 77%. The update gives no absolute base or measurement window. It supports a change in use, but cannot establish commercial scale.[15]
Post-Mortem
Existing enthusiasts hid the adoption problem
The founder update reproduced on Ho’s public profile describes a discovery problem. Current users valued social coworking, while prospective technical leads wanted better calls they already held. This account was available as indexed profile content; the original post’s exact date was not established.[15]
In the Engineering Founders episode notes, Embiricos says: “the metrics don't lie and they're not good.” The interview records the founders’ need for a larger change.[5]
The mechanism was a selection problem. People already using a virtual office could explain its appeal, while people who had not adopted one still needed a reason to change their day. Adding features for enthusiasts could deepen affection without fixing acquisition.
The pivot addressed that problem by improving calls already taking place. Higher screen-sharing use provides directional support. The missing retention and revenue figures leave the strength of the new business unresolved.
Native craft created value and narrowed the audience
Multi’s Mac integration let shared applications feel familiar. It also required specialist engineers and operating-system-specific work. Wood’s account explicitly describes Mac hiring as difficult and the choice as a trade-off. The same investment could improve experience while limiting how easily a mixed-device team adopted the product.[4]
This is an inference about adoption constraints, not a verified cause of the acquisition. The engineering evidence also supports the counter-explanation: specialized desktop expertise was valuable even if a standalone company was not the best home for it.
Reliability depended on system boundaries
Ho documented a 2022 outage in which database retries repeated Slack calls inside transactions. Rate limiting then prolonged transactions and caused further retries. Disabling the notifications also restored server response. The lesson concerns coupled systems: a secondary integration could impair the primary experience.[19]
That incident does not establish chronic unreliability or explain the exit. It shows the operational work behind a product marketed as lightweight.
The customer outcome differed from the team outcome
The official notice ended new signups, gave existing teams an export path, and announced deletion of user data. Greylock now marks Multi as acquired. The team’s destination was confirmed; transaction terms and standalone profitability were not.[1][20]
A strong alternative to a failure narrative is that OpenAI offered a larger setting for the team’s desktop work. The available evidence supports that possibility without proving why the founders accepted. It also does not establish that later ChatGPT features came directly from Multi’s code.
Key Lessons
- Multi’s pivot replaced a promise of more collaboration with a better pairing session. Evaluate the first useful task for a new customer separately from enthusiasm among existing users.
- Remotion’s launch numbers and Multi’s usage changes measure different things. Track acquisition, engagement, retention, and payment separately before calling a pivot successful.
- Native integration improved the product’s character, while imposing platform and hiring constraints. Treat those costs as part of the distribution decision.
- Multi provided an export window when the team moved. An acquisition can create opportunity for builders while imposing migration work on customers.
Sources
- Multi is joining OpenAI
- Talk Live, Type Less: Our Investment in Remotion
- Moving to SwiftUI from macOS Cocoa or iOS Cocoa Touch
- Why Remotion is a native macOS app (not Electron)
- Engineering Founders: Lessons from building in public
- Remotion raises $13M
- How to Launch on Product Hunt - Our Learnings
- Multi changelog
- Renaming Remotion to Multi
- Launching Multi: Multiplayer Collaboration for macOS
- Making Illegible, Slow WebRTC Screenshare Legible and Fast
- Designing Usefully Unintelligent AI Summaries
- Multi product site
- Building a macOS remote control engine
- Charley Ho: founder updates and shared Alexander Embiricos posts
- Tuple
- Apple: Request or give remote control in FaceTime on Mac
- Multi pricing
- Move fast to unbreak things
- Multi: Greylock portfolio