Opportunities › AI Model Inference Platform for Drug Discovery › Proposed app
Benchrun
Run protein and drug discovery AI models from a browser, with no GPU setup
- Score
- 36
- confidence 51
- Evidence
- 6
- items on the page
- Rivals
- 2
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2022
- Emerging market · AI estimate
ScreensConcept
Desktop app: Scientists submit computational models from simple forms and monitor job queues during long research sessions at their workstations. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
New run
Ready to run- ModelStructure predictor
About this appAI proposal
Benchrun lets bench scientists run computational biology models from a simple web form. Pick a model, upload a sequence or structure, set a few options and get results. The platform handles GPUs, containers and queues, so nobody has to write or run scripts on a scientist's behalf.
Chain models into pipelines, for example a structure prediction step followed by a scoring step, and save them for the team. Data science leads can add their own models once, package them as a form, and let colleagues run them without further help.
It is built for biotechs and research labs where a few technical people serve many scientists. Each run is logged with its inputs, model version and cost, so results can be repeated and reviewed.
- Tagline
- Run the model. Skip the cluster.
- Voice
- ClearPreciseCalm
- Why this name
- A bench is where scientists work, and a run is one job. Together they say models are run at the bench, by the scientist.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Model catalog with plain forms
Each model has a guided form with sensible defaults, so non-technical scientists can start a run without code or command lines.
No member yet
Pipelines of chained models
Connect the output of one model to the input of the next. Save the pipeline and rerun it on new batches of inputs.
No member yet
Bring your own model
Upload a container or weights, define inputs and outputs once, and Benchrun turns it into a form for the whole team.
No member yet
Automatic compute scaling
Jobs queue and scale across GPUs on their own. Users see status and estimated time, not infrastructure.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#0F4C5CDeep Teal · primary#F6F4EELab Paper · surface#14232AInk · text#E09F3EAmber · accent#7FB7A8Sea Glass · data
Typography
Aa
Benchrun
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Run the model. Skip the cluster.
PricingAI proposal
The material gives no competitor prices. These tiers are a starting point to test, with usage-based compute as the monetization concept suggests.
Starter
Free
- Public model catalog
- Pay per compute hour
- 1 user
- Run history
Team
$400 a month
- Up to 10 users
- Pipelines and shared workspaces
- Project budgets
- Compute billed at usage
Enterprise
Contact us
- Custom models
- Priority GPU access
- Single sign-on
- Dedicated support
Would you pay Team · $400 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
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
- Users invite others. A scientist who builds a pipeline invites colleagues and collaborators to run it and view results in the same workspace. Shared result links let outside partners see outputs.
- Do they become customers?
- Invited colleagues in the same company can become users on a paid team plan. Outside viewers of a result link may not buy. Test whether invited collaborators start their own workspaces.
- First customers
- Data science leads at biotechs with 20 to 200 staff who run models for other scientists on in-house GPUs or shared clusters.
- Where to reach them
- Computational biology communities: Share worked examples of pipelines in protein design forums and open-source model communities.
- Direct outreach: Contact data science directors on LinkedIn and offer a free pilot replacing one script they run weekly.
- Hacker News launch: Post a demo and honest comparison, since the topic drew interest there.
- First milestone
- Five biotech teams each run a weekly pipeline for a month and three agree to a paid Team plan.
Why people already pay for this
- VerifiedHacker News post received 85 points and 20 comments
- Public claimLarge biopharma companies face workflow bottlenecks where data science directors spend significant time running computational scripts for researchers
- Public claimThe platform serves a substantial customer base across biopharma
- Public claimCompanies actively replace internal infrastructure with the service
- Public claim
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 computeOn-demand GPU cloud with job schedulerModel runs need GPUs that scale up and down per job, which Vercel and Supabase do not provide.
- This app · Container runtimeSandboxed container executionCustom models arrive as containers and must run isolated from other customers.
RisksAI inference
- CommercialThe material says one existing service already serves many biopharma customers. Winning means beating it on a narrow niche, price or model coverage.
- DataPharma users will worry about proprietary sequences and structures. Buyers may ask for isolation, audit logs and private deployment before trial.
- TechnicalPackaging many open models reliably, with pinned versions and GPU memory limits, is ongoing work.
- Intellectual propertySome models carry licences that limit commercial use or hosting. Each one needs checking before it enters the catalog.
- CommercialGPU costs may squeeze margins if usage pricing does not cover idle capacity and queue peaks.
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.