Skip to content

Building with AI

If you wish, AI takes over the implementation

Anything you build in nexxcite in low code in the editor, AI can also implement for you if you wish. You describe what the process needs in your own words. The AI uses the same building blocks as the editor and creates the result as a draft that your team reviews and approves.

  • Every part of the platform editable by AI
  • Every change goes into a draft first
  • Can be undone at any time
  • Your own AI keys
Book a live demo Architecture and governance

The cycle

Four steps to approval by your team

Describe What the process needs In full sentences, in the language of your business team.
Draft The AI creates it Data model, forms, workflows, rules and roles, using the platform's building blocks.
Review Change by change Every proposal is visible in the draft. Your team sees what is new and what changes.
Approve Publish together The draft is published as a whole and can be undone as a whole.

AI as an option

If you wish, AI takes over the implementation

You describe in your own words what the process needs. AI implements it with the same building blocks you use in the editor and places every change in a draft. You review the result, continue in the editor or discard it. Only what a person approves is published.

Draft
Draft
MB

Set up complaint handling: sales records the complaint, quality reviews it, an 8D report is required for recurring defects.

  1. + Data model Complaint
  2. + Form Record complaint
  3. + Workflow Review and 8D escalation
  4. + Rule Detect recurring defects
  5. + Role Quality assurance
  6. + Dashboard Complaints by cause
6 changes in the draft · none of them is live Review changes Approve
Draft
MB

From now on, every order above €25,000 also needs approval from the management board.

  1. ~ Rule Approval level by amount
  2. ~ Workflow Approval with escalation
  3. + Role Management · Approval
  4. ~ Notification Approval requested
4 changes · requests in progress stay untouched Review changes Approve
Draft
MB

Give the customer portal a light theme in our brand colours and our logo. Customers only see their own requests.

  1. + Theme Customer portal light
  2. + Navigation Portal menu
  3. ~ Role Customer (external)
  4. + Access rule Own requests only
4 changes · apply to the customer role only Review changes Approve

Reversible at any time

Building blocks, not source code

AI works with the same building blocks your team sees in the editor. No program code is created that someone has to understand and maintain later.

Every change in a draft

What AI proposes is collected in a draft, visible change by change. Nothing is published until someone approves it.

Only with the user's permissions

The assistant acts within the permissions of its user. If you can't change roles, you can't change them through AI either.

Undo changes

Every draft knows the state before its first change. One click takes you fully back to that state.

Traceable origin

Each change records which AI request it came from. So you can still see later why something was set up the way it is.

You choose the AI model

AI runs through your own account with the AI provider, or with an open model in your own data centre if you prefer.

The difference

The AI works with the platform's building blocks

AI that generates program code and the AI in nexxcite seem similarly fast at the start. The difference shows once applications have to be maintained and changed over time.

AI generates code

Every application is separate software

Every generated application has its own structure, its own dependencies and its own way of handling security.

  • Every application needs review, testing, installation and operation
  • Permission checks are implemented anew for each application, or missing
  • The same requirement produces different results each time
  • Changes mean generating again, with an uncertain outcome
  • Knowledge sits in individual people's chat histories
AI configures nexxcite

The result stays part of the platform

The AI works inside the platform's model. It produces the same thing your team builds in the editor, and the result stays readable, changeable and consistent.

  • Your team sees the result in the editor and can keep working on it there
  • Everything goes through one central permission check
  • The same building blocks and patterns in every application
  • Changes go into a draft, are reviewed and approved
  • Knowledge sits in the model, visible to the whole team

Reach

What the AI can create for you

The AI can edit every part of the configuration, including workflows, rules and roles. That is how a description becomes a complete draft for a process.

Whole processes from one description

“Set up complaint handling with a quality review and an 8D report for recurring defects.” The AI creates the data, forms, workflow, roles and reporting and presents them to you as one coherent draft.

  • Data model
  • Forms
  • Workflow
  • Roles
  • Dashboard

Data model

Data objects, fields, picklists and relationships, added to your existing data model.

Forms and lists

Forms with required fields and validation, lists with filters, dashboards with key figures.

Workflows and rules

Approval levels, deadlines, reminders, escalations and calculations, described the way you would explain them in a meeting.

Roles and access rules

“Team leads only see their own cost centre.” The AI turns this into an access rule that takes effect.

Themes and navigation

Colours, logo, menus and start page of an area, in line with your corporate design.

How it works

From a description to a running application

  1. 01

    Describe

    You explain the process as you would to a new colleague: who requests, who reviews, what happens above which amount and who may see what.

  2. 02

    The draft takes shape

    AI creates the required definitions and collects them in a draft. None of it is visible to your users.

    • Created
    • Changed
    • Origin per change
  3. 03

    Try it in the draft

    Whoever has the draft open already sees the application as it will look after approval and can try it out before other users see it.

  4. 04

    Refine

    “The supplier field should be required.” Further instructions go into the same draft. Anything your team prefers to adjust in the editor goes there too.

  5. 05

    Approve

    The draft is published as a whole. From that moment your users work with the new state, without installing a new version or interrupting operations.

  6. 06

    Undo if needed

    The draft knows the state before its first change and can be reset to it completely.

Security

The AI works with its user's permissions

The assistant has no special rights. It acts within the permissions of the person using it. If you are not allowed to change roles, you cannot do so through the AI either. Data that the permission check filters out is not visible to the AI.

Governance

Every change stays traceable

When AI is involved, a single wrong suggestion is rarely the problem. It becomes difficult when nobody knows exactly what was changed, when and why.

In nexxcite, changes are collected in a draft. It groups what belongs together, makes it reviewable and is published as one. For each change, the AI request it came from is recorded.

Drafts can also be exported and imported into another environment. That way you build and test changes in a test system first and then move them to production.

  • Changes are collected in a draft, reviewed and published together
  • The AI request behind each change is recorded
  • The previous state is kept and can be restored at any time
  • Several people can work on the same draft
  • Export and import drafts between test and production
  • The AI has the same rights as its user

Who benefits

What the AI means for three roles

Administrators

Routine work gets done faster: a new field in five forms, one more approval level, a theme for the new site.

  • Less clicking
  • Have configuration explained
  • Easier start for new colleagues

Business teams

Whoever knows the process can describe it and see straight away what it looks like as an application. Requirements can be settled in conversation.

  • Process in your own words
  • Result visible immediately
  • Quick rounds of feedback

IT and compliance

Control stays with you: drafts, approvals, resetting and a central permission check that applies to the AI as well.

  • No uncontrolled one-off solutions
  • Traceable changes
  • Your own models and keys

FAQ

What customers ask first

What if the AI builds something wrong?

You see it in the draft before other users do. You correct it by instruction or in the editor, or discard the draft. Only what has been approved is published.

Which language models are supported?

Azure OpenAI, Anthropic and OpenAI-compatible endpoints. These also let you use open models that you run in your own environment. You use your own API keys.

What does the AI cost on top?

Nothing from us, as there is no separate AI licence. You pay your model provider directly for usage, or you run the model yourself.

Does our data leave the organisation?

Only if you decide so. With an open model in your own environment, everything stays internal. In addition, the AI only sees what the respective user is allowed to see.

Do we still need our own know-how?

Yes, and that is intended. The AI takes work off your hands, but your team decides and understands what it approves. We support the first process and train your team along the way. The second one you build yourselves.

Can the AI explain existing configuration too?

Yes. The AI answers a question such as “Why does this request go to the head of department?” from the actual rules. That helps new colleagues and every handover.

Live demo

See how the AI implements your process

In the demo you describe a process from your organisation and see it become a draft that you review, change and approve.

Book a live demo Areas and themes
  • A process from your organisation
  • Built live, in the editor or with AI
  • No obligation