A fund's only lever is picking. Ours is manufacturing — automated demand discovery, near-zero-cost build, operators on tap, and capital, running as one system. Here's why the structure wins, what it has already returned, and the ecosystem we built to scale it.
Venture capital is an extraordinary asset class with one permanent constraint: it can only select from what walks through the door, and it must pay the market price for whatever it selects. Every other weakness follows from those two facts.
| Venture capital | Persist — the factory | |
|---|---|---|
| Where shots come from | Inbound deal flow you don't control | Demand discovery you operate |
| Price of entry | Market price at the round | Build cost — $50K on Bump.fm |
| Ownership at first mark | ~10–20%, diluting each round | 33% retained post-raise |
| Shots per $2M | One seed check | ~40 builds |
| Time to know | 5–10 years to an exit signal | Weeks — instrumented operating data |
| Finding the operator | A search, run once per company | 50,000/yr funnel, 9-day median to offer |
| What compounds | Brand and network | Brand, network, and the machine itself |
We are not competing with venture capital for deals. 50+ VC firms already post theses into our funnel. The factory sits upstream of the fund — we manufacture the company a fund wants to buy, and we own a third of it before the first round prices.
A $50K position is a check no fund is structured to write. That is precisely the gap the factory occupies.
The argument above isn't a projection. It's a description of two things that already happened.
Running SongGPT — a live AI music generator — taught us something about music the industry hadn't priced. We built against it for $50,000, placed an operator out of our own talent funnel, and retained a third of the company through a raise at $30M.
"The music industry has a distribution layer (streaming)… but no capital layer. Fans create hits, but capture zero upside. Artists go viral, but barely monetize momentum. Platforms own discovery — and the economics."
Before there was a factory, there was a question: which app wrapper is worth building? Answering it by hand produced a ranked consumer app with more than a million organic users and no paid acquisition. That result is the reason the rest of this page exists — it proved demand discovery was the highest-value step in the chain, so we automated it. That automation is Langisia, now running as a factory with a growing set of performing assets.
Position value reflects the most recent priced round and is an unrealised mark, not a realised return.
Two wins, two lessons: demand discovery is the highest-leverage step, and the right operator is what converts a validated asset into a priced company. Everything below was built to do both of those on repeat.
A portfolio is a list. A factory is a system where each asset lowers the cost of the next one. These are the transfers already running.
Insight from running a live AI music product surfaced an unpriced gap in the same industry. A standalone studio would have had to buy that insight through research. We got it as a by-product of already operating.
The manual search that produced a 1M+ user app is now automated. Langisia finds real demand, builds native apps and prepares them for launch — turning one hit into a factory of shots.
Bump.fm's value was unlocked by the right operator. Most studios go to market to find one. We run a funnel of 50,000+ builders a year with a 9-day median to offer — operator placement is inventory, not a search.
50+ verified VC firms post theses with capital and a milestone attached. The hardest question in venture — what should we build? — becomes an inbound queue underwritten by someone else's balance sheet.
Who shipped, who stalled, where the week went — across every portfolio company at once. Selection stops being instinct and becomes a measurement. This is the layer that makes 198× repeatable instead of anecdotal.
A frontier startup society next to Stanford. Physical density is the cheapest way to meet the highest-agency people early, before anyone else is bidding for them.
Six layers, all live. Persist is the venture factory; Colabio is the factory floor. Click any layer to open it.
120+ curated roles, salary and funding, 9-day median to offer. Compete, ship something real, and run a company.
Startupathon →Define the outcome and the capital. Founders build the milestone; you fund on delivery. 50+ verified firms already do.
RequestForStartups →Capital, structure and operators behind the bet you already made. Cohort 2026 is open.
Apply to Persist Foundry →The workspace our portfolio operates in. Who shipped, who stalled, where the week went. Free plan, web and desktop.
Colabio →Most funds tell you what they'd back.
We show you what we built.