Turns space into place.
A recording is most meaningful at zero distance from its origin. Every mainstream platform optimizes for how far content travels. We optimize for return, not reach.
Throughout, a ™ is a Time Machine: a single recording left at a single coordinate.
Time Machine lets you record — audio, video, photo, writing — and leave it in the exact place it was made. It can only ever be experienced by physically returning to that place. Nothing on Time Machine can be viewed remotely, forwarded, or fed. Content made here is never consumable there, by design.
Leaving is a single, intentional act: recording and placing happen at once, because the coordinates where you record are the only coordinates where the recording will ever exist. Leaving doesn't publish a recording. It plants it — geolocation-gated to the exact spot, invisible and inaccessible anywhere else on earth. Returning is the key. Relive is what happens when presence and place align.
| Mode | Who unlocks it | Depends on density? |
|---|---|---|
| Private | Yourself, on return | No |
| One-to-One | One specific person, whether or not they've ever stood there | No |
| Public | Any user willing to travel to a curated place | Yes — which is why it stays small and curated |
Public extends the mechanic without breaking it: Grizzly Peak, an arts district in Kibera, a street in Kathmandu. The audience can be many; the distance is still always zero.
Every mainstream platform is built to answer one question: how far can this content travel? Reach is the value metric. Time Machine is built on the opposite premise — a recording is most meaningful at zero distance from its origin. We optimize for return, not reach.
A space is just coordinates. A place is a space someone has lived in.
Standing where a recording was made, you aren't scrolling past someone's decontextualized present. You're standing inside your own past, and it hits like time travel.
The moment you let content made here be consumed there, you've built a distribution platform and joined a war you can't win. Our refusal to distribute is the product. Everyone else turns place into content. We turn space into place.
Live on iOS and web. No revenue, no paying users yet. Roughly 20 friends on the app and web client, ~40 users through intimate events (Hilgard Avenue, Cafe Time Machine). Broader reach through physical installations — Grizzly Peak, Phi Psi Surfrider, Memorial Glade, Kibera — where participants left written and audio input off-app.
Stack. Native iOS (Swift/SwiftUI) with an on-device audio and ML pipeline (SoundAnalysis/Core ML, vDSP, Speech) for key/tempo, species ID, speaker, mood, scene, and transcription. Firebase backend (phone-first auth, Firestore, Storage, Node Cloud Functions), web client, and a tmkit SDK. Claude handles server-side curation and moderation. Built almost entirely with Claude Code.
Roadmap. Fully spatial output — media that follows the listener across space, preserving tempo, pitch, and timbre relative to how it was recorded. Beyond that, place-conditioned models: models that live in one place, trained on data that exists only there.
The first market is a university, not a city — and the first is Berkeley. Campuses are the highest-ritual, highest-density, cheapest, most creative-supply-rich places on earth, and the founder is already embedded there. It's Facebook's move — launch one campus, prove it, spread campus to campus — and it fits Time Machine even better.
Corridor #1: the Berkeley campus. One saturated, visible corridor, seeded by people the founder already knows, proving the density-and-return curve before a dollar goes to a new market.
A deliberate remix of go-to-market moves that already worked, each mapped to a specific Time Machine mechanic.
| Company | Their move | Time Machine's version |
|---|---|---|
| Gated campus-by-campus rollout plus scarcity | Launch Berkeley; campuses earn in and spread via graduates. Scarcity as the growth engine. | |
| Uber | Independent contractors, a repeatable local launcher, dispatch and booking; own no cars | Residents, hosts, and bureaus are independent community, not payroll; a repeatable campus-launch kit; artists book a stage and hosts book a party at a ™ in-app. We own the platform, not the supply. |
| Airbnb | A marketplace of hosts plus a trust and standards layer | Venue and resident hosts supply the spaces; certification is the trust layer that makes a stranger's spot legible. |
| OpenTable | Send venues traffic; the venue benefits and pays for the flow | Venues get foot traffic and climb the earn-your-machine levels; the deal scales with what we deliver. |
| McDonald's | A real-estate company behind the franchise | Own the ground under the proven nodes over time — the ownership ladder (§11). |
| Kirkland (Costco) | Own the high-margin private label while shelving competitors | Own our flagships, festival, curated media, and artist stakes — while the indie café down the street still feeds us reach and data. |
| Tabasco | Grew by making the bottle's hole bigger — more use per user | Design for more contribution per person: rotating prompts, more places, streaks, the machine-upgrade game, "play the space" that rewards giving more. Lower the activation energy; raise the frequency. |
| 18Birdies | Club-by-club was heavy; a universal subscription scaled better | A caution, not a template: use clubs and campuses to distribute for free, and watch the operational cost of per-venue deals. A universal subscription may be the durable backbone (§10). |
Two layers, and the machine is often the seed, not the reward.
The digital layer — private and peer-to-peer — works anywhere with just the app. No hardware, no permit; a bare geofence is enough, and these ™s radiate to any coordinate.
The public stream lives only at the machines and ™ events. Public is a locus thing — a physical machine or an event around one, never a bare geofence. Most machines are simply clearly-marked public objects — a payphone, a mailbox, a booth — standing in public, always on. Events cluster around some machines; the machine, not the event, is the constant.
So it isn't a clean earn-your-way-up-to-a-machine ladder. In a place with no users you often lead with a machine or an event to put Time Machine in front of people and start the activity. The first machine in a new market is a deliberate seeding investment, like the campus itself. Capital follows proof for the bigger machines thereafter.
| Level | Physical form | Streams it carries | How it arrives |
|---|---|---|---|
| L0 | Digital only (geofence + app) | Private + peer-to-peer | Free — anywhere someone has the app |
| L1 | Removable marker (sticker/QR/NFC) | Private + peer-to-peer | A light cue at a spot people use |
| L2 | Certified hexagon plaque | Private + peer-to-peer | Certification once a spot is active |
| L3 | Payphone / mailbox kiosk | + Public stream (the locus) | Seeded to start a place, or earned where density proves it |
| L4 | Flagship (booth / hex box / dome) | + Public, full spectacle | A marquee node — seeded or co-owned |
The digital layer radiates; the public stream stays put and curated. Openness drives density; the machine anchors and curates the public show.
What the physical machine actually is. A clearly-marked public object for recording the present. Everything can be done on the phone — the machine is never required — but standing in public it does three things a phone can't: it is a billboard marking the spot as a Time Machine; it plays back the place's output, embodying what the space holds; and it is a communal, tactile station people gather around to record the moment. The phone is the reach; the machine is the always-there beacon, vessel, and invitation.
Seeded — placed to start a fresh market. Earned — organic aliveness across the streams justifies one where users already are. Summoned — a community runs a collect-campaign to bring one to their spot. The seed is an investment; earned and summoned follow proof and demand.
And sometimes a hunch. Not every machine waits for proven demand — sometimes you put one up on a well-reasoned guess and see if people take to it. Keep these bets cheap and few.
Two guardrails. Keep the collect game separate from the genuine public stream — it earns the machine but is never what plays on it. And keep it product-agnostic: safe, on-vibe partners (a local roaster, a record store, event tickets), nothing regulated.
A ™ needs no permit if it has no footprint. A pure geofence occupies nothing, so most public space is free the moment you stop assuming a Time Machine must be an object. Only permanent hardware or organized crowds trigger permits. Use the lightest lever; escalate only where the place earns it.
| Lever | Use for | Instrument |
|---|---|---|
| 0 · Digital | All public land | Geofence + app registry. No permission. |
| 1 · Marker | Sidewalks, poles, quads, vacant lots | Removable, non-damaging sticker / QR / NFC |
| 2 · Host yes | Cafés, bars, galleries, shops, club houses | One owner's yes — a night and foot traffic in exchange |
| 3 · Institutional | Libraries, museums, the university itself | Program office / MOU; brings power and audience |
| 4 · Municipal | Parks, squares, cemeteries (hardware or crowds) | Permit / public-art program / a BID partner |
| 5 · Piggyback | Any permitted existing event or club night | Join as a participant — inherit permit and crowd |
On a campus most of this collapses to Levers 2 and 5: a club's yes and an existing party. Permitting never scales centrally — the local resident secures their own spot, and the digital-first default means most ™s need no permission at all.
Not one silver bullet, but a sequence of streams, seeded locally and scaled through the substrate. Per-place economics funds early density and proves the model; the platform layer is the venture-scale upside.
The 18Birdies note. For their golf app, club-by-club proved operationally heavy and a universal subscription scaled better. Not a verdict for us — our per-place events are also marketing and supply — but a caution: watch the cost of many small venue deals, and treat a universal subscription as the likely durable backbone if the per-place hustle gets heavy.
Never sell user data. Presence-gated memory is the brand and the moat.
We won't claim we can size the substrate line credibly yet. What we can say is that it never requires crossing the one line the whole company is built on: turning someone's presence into something we sell to anyone other than the person who was standing there.
Asset-light to prove, owned to profit. Uber owns no cars until it knows the route; McDonald's owns the real estate under the franchise; Kirkland owns the label while shelving the rival. Time Machine climbs the same rungs — and the café down the street still feeds you reach after you own the flagship.
Own only what you've proven with usage. Earn the right to own, the same way a place earns its machine.
Asset-light (own the platform, not the supply), community-powered (free artists for a stage), AI-leveraged (a tiny team). Money goes to leverage and earned hardware — not headcount, not paid artists, not real estate yet.
What AI runs — this is what keeps the team tiny. It builds and maintains the software (already underway with Claude Code); runs curation and moderation through a per-place model plus the safety filter; applies the house style automatically; drafts editorial; and handles ops — onboarding, triage, party calendar, kit generation, resident support. AI absorbs the costs that force hiring, so headcount grows sublinearly with reach.
| Funding level | Core paid team | The rest is done by |
|---|---|---|
| $50k | 0 new hires — founder plus room-and-board crew | Free student community and AI. Spend goes to tech, one or two machines, and markers. |
| $500k | ~3 — a tech lead, a campus and community lead, a curator-editor | Community, contractors, and AI. Budget reweighted toward software and curation. |
| $2M | ~6–9 — a 2–3 person product pod, two campus launchers, a curator/editorial lead, ops and partnerships | Community and AI at scale; residents, hosts, and artists remain unpaid supply. |
One campus to undeniable aliveness, then let it spread. Each quarter has a gate; don't advance until it's met.
| Quarter | Goal | Do | Gate to advance |
|---|---|---|---|
| Q1 | Prove aliveness in one corridor | Berkeley campus; 5 anchor residents; digital-first plus markers; put ™s at the parties and club nights already happening; ship the prompt deck, house style, and integrated app | ~20 ™s per corridor; students still open the app there two weeks later |
| Q2 | Own the campus | Saturate the quads and key venues; certified plaques (L2) plus first L3 machine; first ticketed premiere; turn on first revenue | App useful daily on campus; first revenue; repeatable event kit |
| Q3 | Own the college town | Spill into Berkeley and Oakland venues; first flagship (L4); open the resident call; stand up bureaus; first sponsor | A known monthly fixture; strong retention; open call oversubscribed |
| Q4 | Prove repeatability | Document the campus-launch playbook and metrics; admit one second campus that asked; begin the tmkit/SDK pilot; raise the scale round | A playbook a stranger could run; economics that fund campus #2 |
Everything templatized so nothing has to be negotiated — keeping transaction cost near zero at ten-thousand scale.
| Kit | What's in it | Tier |
|---|---|---|
| Marker Kit | Hexagon stickers, stencil, QR and NFC tags, a three-step "designate this spot" card, app link | L1 · anyone |
| Prompt Deck | Rotating universal prompts (cards and in-app), house-voice guidelines | All places |
| Resident Kit | Certified plaque, dashboard, party slot, curation tools, revenue-share template, swag | L2 · residents |
| House-Style Kit | The design system, editorial standards, and role briefs — correspondent, filmmaker, writer, musician, photographer | Bureaus |
| Party / Host Kit | "Throw a Time Machine night" guide, signage, prompt cards, app setup, safety basics | Party hosts |
| Machine Spec | Reference designs for the L3 kiosk and L4 flagship; power, connectivity, upkeep | L3–L4 |
| Campus Launch Kit | The full playbook, brand assets, standards, ritual calendar, access field guide, metrics dashboard | New campuses |
Ignore total installs — the only vanity metric. Track aliveness and raise on the curve.
Not the near-term driver; a mode we grow into. Prove the model on Berkeley first. We won't have to push: once one campus is undeniable, the next ones ask. A light apply-to-earn-in gate lets a campus prove it's ready — a champion, an artist roster, committed venues, a scene — which pre-seeds the supply and siting a launch needs.
Keep it warm and accessible, never velvet-rope. The gate is about readiness and density, not status; Time Machine is populist. And it pays: earning in makes college fun, piggybacks existing clubs, and turns campus pride into free marketing. The applicant queue is itself demand proof for investors.
It's geocaching with a party to fill each bucket — and we sell the party. Geocaching proves a global network of millions of presence-gated, community-placed sites works with almost no central operations. We add the media, the memory, the artists, the curation, and the ritual.
Every meaningful place on earth holds its own memory, authored by the people who were there, gated by presence — and a traveler anywhere sees a hexagon and knows the ground beneath them is remembering.