No engineers to hire or keep
No recruiting, no onboarding, no salary floor that exists whether or not there is work this month, and no key-person risk when someone leaves in year two.
Efficient
You pay for what you use. Not for a team.
The expensive part of software has never been the building. It is the payroll that has to exist afterwards.
A cost that tracks your actual usage, and no fixed headcount underneath it.
Compare the options honestly. A technical co-founder costs a permanent slice of the company, before your first customer. An agency costs six figures and six months, and hands back something nobody is left to maintain. An in-house team is two or three salaries, benefits and management attention, sustained forever. Against those, a licence plus the resources your application genuinely consumes is not a discount — it is a different category of commitment.
What you are not paying for
No platform to maintain
No build pipeline to look after, no monitoring stack to configure, no upgrade weekend. The infrastructure work exists — it is simply not yours.
No equity for a technical partner
The most expensive way to buy engineering is to pay for it in shares of a company you believe will be worth a great deal. You keep the shares.
No six-month wait before you see anything
Time is a cost even when nobody invoices you for it. Two quarters spent building the wrong thing is the single most expensive mistake available to a new company.
How the cost actually works
Two components, and you can see both.
A licence for the platform
One predictable line: the platform, the building, and everything that keeps your application running. For an incubator or accelerator this is one licence covering the whole programme.
The resources your application uses
Compute, storage, bandwidth, email volume and any AI usage — billed as consumed. A quiet application is cheap to run, which matters most in exactly the months when it is quietest.
Usage visible before the invoice
You can see what your application is consuming as it consumes it. No surprise reconciliation at the end of the quarter.
Limits you set
Spending thresholds with alerts, so a runaway job or an abusive script becomes a notification rather than a five-figure bill.
Efficient by construction
The cheapest infrastructure is the infrastructure you do not run.
Capacity that scales down as well as up
Idle capacity is released. You are not paying overnight for the peak you needed at lunchtime.
Shared foundations, isolated data
The common machinery — monitoring, backups, pipelines, security tooling — is built once and maintained across every application, while your data stays separated. That is what makes this affordable at all.
Waste found and removed
Oversized resources, storage nobody reads and jobs that run more often than anything needs are cleaned up as part of running the thing, not as a project you commission.
AI usage sized to the job
Model choice matched to the task rather than defaulting to the most expensive option, with caching for repeated work. AI cost is the line most likely to surprise people, so it is watched deliberately.
The specifics
The cost model
Exact figures depend on what you are running. The shape does not.
| Item | What you get |
|---|---|
| Platform licence | Fixed and predictable — per company, or per programme for partners |
| Compute | Metered by use, scales to near-zero when idle |
| Storage | Metered per gigabyte, database and files |
| Bandwidth | Metered, CDN-served |
| Metered per message sent | |
| AI usage | Metered by tokens, with model choice matched to the task |
| Environments | Staging included; it does not carry a second licence |
| Monitoring, backups, security | Included — not a premium tier |
| Support | Included |
| Spend controls | Thresholds and alerts you set |
Questions people ask
What does it actually cost?
A licence plus the resources your application consumes. The licence depends on what you are building and, for a partner programme, on the size of the cohort. The resource line is genuinely usage-based, which for most early applications is a small number. See Pricing for the shape, and a call for the figure.
Why is the resource cost separate from the licence?
Because it is a cost you would carry no matter who built your application, and bundling it into a flat fee would mean charging quiet customers for busy ones. Keeping it separate keeps it honest and keeps it visible.
What happens if we stop?
You get your data out, in a usable format. Being able to leave is a property of a service worth staying with.
Is this cheaper than hiring a developer?
For one application, substantially — you are not carrying a salary through the months when there is nothing to build. If you reach the point of needing a permanent in-house team, that is a good problem and we will tell you when we think you have reached it.