An operational workspace
Jobs, statuses, documents and decisions in one view. For example, a request reaches the right person while AI prepares a summary for review. We agree user roles and access to cases before implementation.
Custom AI systems. Your data, workflows and decisions in one place.
I build custom systems that bring together a working interface, data from your tools and purpose-built AI features. You can check a case, review a recommendation and approve the next step without searching across several applications.
I start with one workflow: managing jobs, reporting, reviewing inventory or working with documents. The system is designed around that task. AI helps interpret content and prepare proposals, while calculations and operating rules are explicitly defined.
Discuss your workflow
What could your system help you manage?
Reporting and analysis
Data from a store, warehouse or files feeds a shared report. Rules calculate values; an AI module can explain changes and flag issues for review. Incomplete data must remain visible before conclusions are drawn.
Knowledge and documents
An interface helps users find information in approved materials, compare documents or draft answers with source references. We check document freshness and define when the system should ask for clarification.
Decision support
I combine rules, priorities and AI analysis into a list of cases that need attention, such as inventory reviews or marketing plans. The system prepares material for a decision. Any autonomous actions are agreed separately.
The interface, integrations and AI each have a role
The interface presents the work to the user. Integrations supply data and pass on approved changes. Rules perform calculations, while AI helps interpret content or draft recommendations. This makes it possible to distinguish source data, calculated results and model suggestions.
Before quoting, I check APIs, permissions and restrictions in your existing tools. We also agree local or server-based operation, activity history, data exports and third-party costs. Some integrations need additional setup or licences.
Examples from my portfolio
These are specific implementations, not fixed feature packages. I define your project scope separately. Client data and identities remain private.
Advanced e-commerce reporting
I extended a client’s system with a report combining sales, product costs, marketplace fees, advertising, delivery and returns. Filters for period, channel and account help focus the analysis. Missing costs remain visible rather than being treated as zero. This illustrates integration and calculations that do not need AI at every step.
Explore this project
AI Inventory Monitor
Local rules calculate product inventory conditions. New or materially changed cases go to AI to help explain recommendations. I added prioritisation, analysis limits and model cost controls. The user makes purchasing decisions; the module does not place orders.
Explore this project
AI Marketing Centre
I built a centre with six agent roles, a shared plan, reports and approvals. Analyses are started manually, with a Marketing Manager role organising the findings. The owner retains control over publication and budgets.
Explore this projectImages illustrate the scope, not a live client interface.
From an enquiry to a completed order.
Watch an example of submitting an enquiry, calculating a quote, accepting it and requesting design changes. It shows how customer collaboration could work in a single dashboard.
Try the same example in the linked demo. It uses fictional data, and changes disappear when you refresh the page.
Try the panelControls for day-to-day operation
Data and access
We agree which system is the authoritative source, which permissions are needed and who can approve changes. Reading and writing data require separate permissions. Trial materials should exclude unnecessary confidential information.
Failures and repeated runs
We define what happens when a connection fails, a duplicate appears or data is missing. Running a task again should not unintentionally duplicate an operation. I test retries and any available reversal procedure for the agreed actions.
AI and decisions
We distinguish proposals for review from actions the software may carry out. AI can make mistakes. Sending messages, changing prices or placing orders requires agreed rules and permissions.
Ongoing operation
Before launch, we agree responsibility for hosting, backups, updates and fault reporting. Support hours, response times and further development are set out in the proposal.
From one workflow to a working solution
Discuss a real example
You show me the current workflow, data sources and desired result. We agree the initial scope, acceptance criteria and maximum billable hours.
Review an initial version
I prepare the selected workflow or interface using agreed examples. You check whether the layout and output suit your work before I build further features.
Test and launch
I check valid inputs, missing data, connection failures and permissions. Writing to your existing systems starts only within the agreed scope after testing.
Handover and support
You receive instructions, agreed access and a walkthrough. Support terms and additional features are agreed separately. You can share project materials and feedback through the client portal.
Pay for actual hours — up to an agreed cap
EUR 35 net per hour
Before work starts, I agree the scope, hourly rate and maximum billable hours with you. If I finish sooner, you pay less. If the agreed scope takes longer, I absorb the extra time.
- Agreed cap: 10 hours. Completed in 6 hours: you pay for 6 hours — EUR 210 net.
- Agreed cap: 10 hours. Completed in 12 hours: you pay for 10 hours — EUR 350 net.
Additional tasks outside the agreed scope require a separate quote and your approval before I start them. Extra time needed to complete the original scope does not increase your bill.
You receive a summary of the time worked. The cap covers my fee for the agreed scope. AI usage, hosting, hardware and paid tools are separate costs, which I explain and agree with you before purchase or activation.
Discuss scope and budgetBefore you get started
Do I need to replace my existing software?
Not necessarily. I can build a separate interface or module connected to your current tools. I first check their integration options and the available data access.
Does every module need AI?
No. Records, statuses, calculations and repeatable rules can run without a model. I add AI where it has a defined purpose and its usefulness can be tested.
Will I receive the source code and rights?
After payment, you receive the source code and economic copyright in the software created for you, as specified in the agreement. External libraries and services remain subject to their own licences.
Can I start with a small scope?
Yes. We agree an initial workflow and acceptance criteria. Further modules can follow after review, with a separately approved scope and maximum billable hours.
Show me how you handle the task today
Describe the task, applications, frequency and result you need. Explain who will use the solution and which decisions need approval. A description or anonymised example is enough for an initial discussion. Do not send passwords or API keys.
Discuss your workflow