Opportunities › Crowdfunded ML experiment execution and tracking › Proposed app
Runpot
Pool small sponsorships to run speculative ML experiments, in public
- Score
- 46
- confidence 51
- Evidence
- 8
- items on the page
- Rivals
- 0
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop and mobile
- PWA: on computers and phones alike
- Market since
- 2012
- Mature market · Verified
ScreensConcept
Desktop and mobile: Researchers propose experiments and review setups at a desk, while sponsors approve pooled sponsorships on mobile, and both monitor live runs across devices. Sample screens on a computer and a phone, in the proposed brand. Their numbers are sample data.
New experiment
Estimate ready$120
About this appAI proposal
Runpot lets an ML researcher post a small, speculative experiment with a fixed compute budget. Sponsors chip in small amounts. When the budget is met, Runpot runs the job on rented cloud GPUs and publishes metrics and artifacts on an open, MLflow-style run page.
It is for researchers and data scientists who can't casually self-fund a run, but whose idea is too uncertain for a grant or investor. Every experiment has a cost cap, so nobody has to approve an open-ended budget.
Sponsors see where their money went: the compute bill, the run log and the result, including negative ones. Cards are charged only when the pot is full, and unspent budget goes back to sponsors.
- Tagline
- Small bets on big questions, funded together and run in the open.
- Voice
- PlainOpenPractical
- Why this name
- A pot is the shared money; a run is one experiment. Two plain words joined, easy to say, and it states what the product does.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Experiment proposals with a cost cap
Describe the hypothesis, repo, GPU type and maximum budget. Runpot estimates the run cost before the proposal goes live.
No member yet
Small sponsorships, pooled
Sponsors give any amount toward the goal. Cards are charged only when the pot is full, so a half-funded idea costs nobody anything.
No member yet
Managed execution
Runpot provisions the GPU, runs the job in a sandbox, stops it at the cap and shuts the machine down.
No member yet
Open run tracking
Metrics, parameters and artifacts show on a public page in MLflow format, updating live while the run goes.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#1F5E4BPine · primary#F6F1E7Paper · surface#1F2A24Ink · text#E0792AEmber · accent#3A6EA5Lake · data
Typography
Aa
Runpot
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Small bets on big questions, funded together and run in the open.
PricingAI proposal
The material gives no competitor prices. These tiers follow the monetization idea in the opportunity (fee on compute, researcher subscriptions) and are a starting point to test.
Sponsor
Free
- Browse and fund public experiments
- Spend ledger and automatic refunds
- Follow runs and results
Researcher
Free
- Post public experiments
- Managed GPU runs with a cost cap
- Platform fee on funds raised
Researcher Pro
$19 a month
- Unlimited experiments
- Priority compute queue
- Private experiments
- Lower platform fee
Lab
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Researchers share the experiment page to ask for funding and later the public result page, with logs and charts, on social media, forums and in papers. Each page shows Runpot and a fund button.
- Do they become customers?
- Visitors can become sponsors, and some sponsors are researchers who could post their own run. Test how many page visitors give, and how many givers later post an experiment.
- First customers
- Independent ML researchers and hobbyist practitioners with an existing audience, running experiments of $50 to $300 in one niche, such as fine-tuning or small-model research.
- Where to reach them
- Hacker News and ML forums: Post a funded experiment as a case study and invite readers of related threads to sponsor it.
- Researcher newsletters and blogs: Offer a few authors a free funded run, with a result page they can embed or link in their posts.
- First milestone
- Ten experiments fully funded by at least five outside sponsors each, with three researchers posting a second one.
Why people already pay for this
- Public claimExperiment.com: 396 projects funded; just over $5 million pledged (as of Jan 2016) (2016)
- Public claimWeights & Biases (CoreWeave Forge) Pro plan: Starts at $60/month, billed monthly (2026-11)
- Public claimExperiment.com: 2012
- Public claimWeights & Biases (CoreWeave Forge) Free plan: $0/mo, 5 GB/mo storage; Pro 100 GB/mo, extra $0.03/GB (2026-11)
- Public claimIndividual ML experiments are cheap but accumulate fast; no one wants to approve speculative budgets.
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · GPU computeCloud GPU rental API (on-demand)Runs sponsored jobs on rented GPUs and shuts them down at the cap.
- This app · Experiment trackingHosted MLflow tracking serverGives each run an open page for metrics, parameters and artifacts.
RisksAI inference
- CommercialThe evidence is public claims that people will sponsor ML experiments. It is not known whether sponsors will pay repeatedly, or whether enough of them exist beyond friends of the researcher.
- TechnicalRunning untrusted code on rented GPUs needs strong sandboxing and hard spend caps. Overruns or abuse such as crypto mining would fall on the platform.
- RegulatoryHolding and refunding pooled pledges may count as crowdfunding or money handling in some regions. Delayed capture through a payments provider needs legal review.
- PlatformGPU supply and prices depend on cloud providers. A price rise or capacity shortage can break the cost estimates shown to sponsors.
- Intellectual propertyOpen results, code and datasets raise licence questions, and sponsors may expect rights. The terms for ownership and any premium sponsor access need to be set early.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.