Atma Sciences (product: Gizmo) was a consumer AI creation platform — a TikTok-style feed of interactive “mini-apps” users could build from a single text prompt, tap/swipe/draw against, and remix. The wedge was collapsing the distance between “idea” and “shareable interactive artifact” for non-technical consumers — Replit/Cursor for people who will never open a code editor, delivered as a For-You feed instead of an IDE.
StoryHouse invested $75K into a $5.5M Seed on a $35M post-money in January 2025, led by First Round Capital (Todd Jackson, $3.5M) with Coatue, Canaan, K5 Global, and Figma Ventures also on the cap table. The thesis was jockey-first: Josh Siegel and three co-founders were a hand-picked ex-Snap consumer team — Josh had led “new products & bets” (Spotlight, consumer AI) at Snap after being one of the first five hires at Looker (acquired for $2.7B per 12/7/24 note). The investment memo (12/7/24 note, recruPxMsVg92Dj7a) called this a “moonshot consumer bet” where the team would either iterate to product-market fit or fail — the tail was a WhatsApp/Instagram-scale outcome. Twelve months in, Gizmo had passed 600K installs and Meta had made verbal offers of $150M then $350M before restructuring the deal into a $40M license-and-release (2026-01-27 note, recvYjlbLa8TNLpJw). Wire of $263,332.84 initial distribution against a $75K check landed 2026-02-01.
Gizmo sat on the consumer edge of the vibe-coding curve — the segment where users are not developers and never will be. That segment is where the largest TAM lever lives; per published industry breakdowns, 63% of vibe-coders are non-devs Web. Meta’s post-acquihire launch of Pocket (June 2026) — a “creative platform for making and sharing gizmos” in a scrollable feed — is direct confirmation the timing thesis was right and the market frame was defensible enough to attract a hyperscaler Web.
| Player | Positioning | Funding / Stage | Edge vs. them |
|---|---|---|---|
| Gizmo (Atma) | Consumer TikTok-style feed of prompt-built interactive mini-apps | Seed, $5.49M (First Round lead) | — |
| Replit | Prompt-to-app builder; large share of users are non-developers | Late-stage; Agent product scaled $10M→$100M ARR in 9 months | Replit ships utility apps in an editor; Gizmo shipped consumer entertainment in a feed |
| Lovable | Non-technical AI app builder | Hit $100M ARR in 8 months | Lovable produces web apps for makers; Gizmo produced disposable playable objects for scrollers |
| Cursor | AI code editor for developers | Category leader for pro devs | Explicitly different audience — developer IDE, not consumer feed |
| Bolt.new | Fast prompt-to-prototype | Vibe-coding challenger | Also builder-oriented; no social/feed distribution layer |
Moat: The differentiated bet was distribution and format — a For-You feed of tappable interactive objects rather than a builder tool — with potential Instagram-for-apps network effects. That format thesis is what Meta ultimately paid $40M to license and rebuild as Pocket Web.
Deal mechanics (per 2026-01-27 portfolio call, recvYjlbLa8TNLpJw): Meta paid $40M for a non-exclusive, perpetual, worldwide IP license and hired all 8 employees. $10M was allocated to employee cash bonuses; $30M for investor distributions (pre-tax). Because it was a licensing transaction with liquidation rather than a stock acquisition, the entity owed corporate income tax with limited NOLs — Josh estimated $8M+ tax leakage, heavier because Atma is domiciled in NYC (~14–15% state/local vs. ~9% in California). Net expected investor distribution: ~$20–22M, implying ~3.5–4.0x on the $5.75M raised (~$5.49M per Company field). StoryHouse’s first distribution of $263,332.84 on a $75K entry pencils to ~3.51x on tranche one, with a 5–10% holdback still pending final wind-down.
Comps for the deal structure: Meta’s license-and-release mirrors Google’s reverse acquihire of Windsurf (July 2025) and the earlier Character AI / Google arrangement, where a hyperscaler hired the core team and licensed IP rather than acquiring the entity — a pattern that has emerged specifically to sidestep antitrust review of AI-talent M&A Web.
Post-signing debrief with Josh on the Meta transaction — how the structure changed from stock acquisition to $40M license-and-release, tax leakage math, and expected investor outcome.
Deal Background & Evolution: Meta initially made verbal acquisition offers at $150M, then $350M, but ultimately changed the structure to a “license and release” transaction rather than a stock acquisition. The final deal is not an acquisition of Atma stock. Instead: Meta pays $40M for a non-exclusive, perpetual, worldwide IP license. All team members receive employment offers at Meta. $10M is allocated to employee cash bonuses. $30M is allocated for investor distributions (before taxes).
Why the Structure Changed: Meta had concerns about optics and regulatory issues around acquiring a small social media startup with minimal hard assets, and the timeline and friction of a full acquisition (estimated 3–4 months). Similar structures have been used recently (e.g., Windsurf, Character AI with Meta/Google). Golden parachute rules were not a binding constraint; founders technically did not have to structure investor payouts this way, but they did because it was the right thing. Josh thinks there will likely be regulation ruling against this, because this loophole is unfair to investors.
Tax Impact & Investor Returns: Because this is a licensing deal with liquidation (not a stock acquisition), the company must pay corporate income taxes. Atma has limited NOLs, resulting in meaningful tax leakage: estimated $8M+ in taxes; NYC taxes are materially higher than California (closer to 14–15% vs. ~9%). After taxes and wind-down costs, expected investor distributions: ~$20–22M net. Implies a ~3.5–4.0x return on the ~$5.75M raised one year ago. Investors will receive an initial liquidation payment this week (~3.5x) and a remaining 5–10% holdback after full wind-down later this year. All investors are receiving the same terms.
Founder / Board Context: Josh acknowledged this outcome is not fair to investors relative to earlier expectations. Original structure would have produced ~10x returns; current structure materially reduced outcomes. Todd (board observer) was looped in throughout: initially supportive; encouraged continuing negotiations as returns fell from ~10x to ~3.5x; believes there is still a 10–20% chance Gizmo could have become a multibillion-dollar company. One co-founder is currently living with parents for example; team opted for deal certainty + Meta platform scale.
Other: Recommended reading — “The Halo Effect” (kwokchain.com, 2025-07-15).
The formal deal call. Josh pitched Gizmo (“Instagram for apps”) with three co-founders from Snap. Term sheet was set with First Round leading; oversubscribed but room made for StoryHouse.
An app to make apps — like a Replit or Cursor but for consumers. Eg. a friend can make a tetris/snake game for a friend, share it, that friend can make some modifications, share it again. “Instagram for apps;” calling them Gizmos. 3 other co-founders, he worked with all of them at Snap.
Financing: Raising $5M on a $35M post-money. Todd Jackson at First Round leading, $3.5M check. Round is oversubscribed but he’ll make room for us.
Special consumer team (Josh Siegel and his handpicked consumer team, first 5 employees at Looker, acquired for $2.7B; 9 years at Snapchat, previously led consumer product management, most recently “new products and bets” including AI, spotlight which have driven majority of Snap value over past 2 years).
Moonshot consumer bet; likely team will iterate iterate iterate until product market fit or fail; likely fail, but if succeed, big WhatsApp/Instagram-type outcome “other” bucket. Top lead investor, FirstRound capital, early consumer investor in Uber, Roblox, HotelTonight, Mint, Warby Parker.
Recommendation — invest $100K. Exceptional team and financing. Next step: send DD questionnaire. One-pager: docsend.com/view/f3j37dxm37sgmbr2.
Email exchange between Miles Bird and Josh Siegel. Josh confirmed he gave notice at Snap the prior week, with 9/20/2024 as his last day. No fundraising plans yet at that point.
Follow-up in 6 weeks (9/23/24). Email response from Josh Siegel.
Josh Siegel (Aug 14, 2024): “Thanks, yeah, I actually just gave notice that I’m leaving Snap last week. My last official day will be 9/20. No plans for fundraising yet but will let you know!”
Miles Bird (Aug 14, 2024): Checked in on wedding congratulations and the timing of the new ed-tech company / when fundraising would kick off.