The one-page decision.
Somebody in your company wants to run a Nuclicore pilot and needs your approval. This page is what you need to decide: what it is, what it costs, what you commit to, how risk is contained, and who is accountable if it goes wrong.
What it is
One workflow, built by an AI engineering department the business operates, in production after four weeks.
What it costs
A fixed price for the four weeks, quoted in writing after a one-hour scoping session. The price depends on the scope, so it is not published.
What you commit to
Four weeks and the fixed price. Nothing beyond that. What is built is yours either way.
Who is accountable
The founder and managing director of Nuclicore runs the engagement himself. You have his name before you sign, and he is in the room for the scoping session and the demo.
What is signed, week by week
Week 1: the brief and the task list
One workflow, success criteria, the systems it talks to. The task list the product manager agent proposes is the contract for the four weeks.
Week 2: nothing
Your team clicks through a working preview and changes its mind cheaply. No commitment is made this week.
Week 3: the integration test
A real record in, a real record out of your system of record, in an isolated test environment. Your operations lead signs off.
Week 4: the production release
Security gate, sign-off, release. A named person in your organisation approves it. It is on record.
How risk is contained
Scope risk
One workflow, not a platform. If the task list has twenty items, the scope is wrong and we cut it in week one.
Operational risk
Nothing touches production until week four. Preview and test have their own databases and their own secrets, and any earlier release can be redeployed from the history in minutes.
Security risk
A security gate on every build, nightly scans, encryption in transit and at rest, EU or German hosting, GDPR data processing agreement with every account.
Vendor risk
The code is standard and yours. Export it at any time. Nuclicore has run software for insurers in production since 2021. If we stop tomorrow, your application does not.
People risk
The pilot needs the business owner of the workflow, one person who knows the data, and an approver. Not a project team.
What happens after the four weeks
You continue
Most customers start the second workflow immediately, because the roles, branding, integration and environments already exist and the second one is faster.
You stop
You export the repository and keep the application, the data and the audit trail. There is no exit fee and nothing to unwind.
Either way
The decision is made on a running application in production, not on a proposal.
What Managing Directors ask
Why not wait until the core system migration is done?
Because it will not be done on time, and when it is, this item will still not be covered by it. The application connects to the core you have now through a defined integration, and is re-pointed to the new core when it exists.
Why not an agency, our own IT, or engineers with a coding agent?
An agency quotes months and a budget for one application, and hands it back as something your team then owns without having built it. Your own IT can build it properly and is booked. Engineers with a coding agent build faster than they used to, and still have to run hosting, environments, secrets, security checks, releases and monitoring themselves. Nuclicore delivers the first application in four weeks at a fixed price, with the product manager, QA, SecOps, DevOps and database work done by agents inside a governed pipeline, and the code is a standard repository your IT can read, scan and export at any time.
Who has done this before us?
Broker Collective: a customer portal in six weeks, now on two brands, followed by an IAM portal, a partner portal, claims management and a purchase flow builder. TAS, part of MRH Trowe: automated renewals replacing PDF forms. Both quoted on the home page.
What is the catch?
Discipline. The pilot works when the workflow is one workflow, when access to the system of record arrives in week one, and when the approver has seen the preview. We hold you to that.
How do we know the software actually works before it goes live?
A task is only handed over once the application installs, builds and runs, and the preview is only reported ready once it answers through its own frontend and its own API. In week three a real record from your system of record goes in and comes out in an isolated test environment, and your operations lead signs that off. The production release in week four is approved by a named person in your organisation and is on record.
Approve the scoping session, not the pilot.
The scoping session is an hour with the founder. You get a brief, a task list and a fixed price. Then you decide.
The four-week pilot, week by week