Skip to content Task
Define value, boundaries and data sources. We identify inputs, connected systems and responsible people. The business process determines what can be automated and where human decisions remain necessary.
DataKG-NET / BUILT AROUND YOU
AI systems.
Clear. Considered. Connected.
AI-assisted features with a defined purpose, controlled data access and human review.
Discuss your project ↗
01Data02Workflow03Approval
KG-NET / AI systems
Assistance with a clear purpose.
AI can organise information, prepare drafts and support staff. We bound the task, define permitted data sources and establish when a person must review. Confidential information is not sent to external models without review.
Planning, design and technology are considered together. Scope and price follow your actual requirements.
01Data02Workflow03Approval
AI systems
How it comes together.
Three perspectives on your project. Scroll through the process or select a step.
Assistance
Build response paths and human oversight. Interfaces receive clear data formats, permissions and error handling. A reviewable workflow explains what happens with missing data, outages or conflicting responses.
WorkflowEvaluation
Review examples, errors and escalation. We test agreed scenarios and document responsibility. Purchases, critical changes and other sensitive actions proceed only after the required human approval.
ApprovalWORKING TOGETHER
Clearly agreed. Clearly delivered.
A defined scope
Before work begins, we agree goals, content, responsibilities and integrations. You know what is included and what needs a separate plan.
Reviewable milestones
Structure, design and function are reviewed in clear stages. Decisions follow your requirements and feedback.
A clear handover
The agreed scope defines access, guidance and an understandable operation and maintenance plan. Ongoing support is agreed explicitly.
AI-assisted features with a defined purpose, controlled data access and human review.
What we build with you
AI can organise information, prepare drafts and support staff. We bound the task, define permitted data sources and establish when a person must review. Confidential information is not sent to external models without review.
- Use case and boundaries
- Data sources and access
- Assistant workflows
- Evaluation and approval
AI features need bounded tasks and assessable outputs. We examine permitted data and how responses will be checked. Uncertain results follow an appropriate review path. Models do not replace permissions or binding business rules. Provider dependency, running costs and possible errors belong in planning so assistants can be used meaningfully within real work processes.
How the project takes shape
Task
Define value, boundaries and data sources.
Assistance
Build response paths and human oversight.
Evaluation
Review examples, errors and escalation.
What makes working with KG-NET clear
We discuss how your business actually works before specifying individual features. Dependencies, responsibilities and the agreed scope are documented. Decisions are discussed using actual pages, tasks or data flows. Questions, materials and approvals are coordinated with an accountable contact. During handover we clarify what your team can maintain and which technical work requires a separate service. You can see what was agreed and where an additional decision is needed.
20 / QUESTIONS & ANSWERS
Decide with clarity.
How are interfaces selected?
We review documentation, permissions, data fields and operational limits. A stable supported API is preferable to unclear access.
Who decides in automated workflows?
Approval points are explicit. Purchases, sensitive changes and payments do not proceed unchecked on an AI decision.
What happens if an API fails?
Errors are recorded and retries limited. Last known data carries a timestamp and missing data is not invented.
How is duplicate processing avoided?
Unique event identifiers and defined state changes help handle retries safely. Typical failure scenarios are tested.
Can barcode and QR workflows be integrated?
Yes, if scanners, identifiers and the target system align. We clarify whether a code represents a product, batch or process.
Where is AI useful?
In bounded tasks such as suggestions or classification whose quality can be assessed. Data access and human review stay explicit.
How are secrets stored?
Tokens and keys belong in protected server-side storage. They are not exposed as browser data or unprotected text fields.
Can the system be extended later?
Modules and documented interfaces limit dependencies. Extensions are checked against the data model, permissions and existing workflows.
How does the project start?
We clarify the audience, existing systems, responsible people and intended outcome, then agree the scope.
What determines the price?
Scope, integrations, content, licences and support are considered separately. You receive an individual offer before delivery starts.
When is the schedule agreed?
After discovery and prioritisation we agree milestones. Approvals, access and external dependencies influence dates.
Who supplies content and access?
We agree which text, images and data you provide. Access should use roles, invitations or limited tokens wherever possible.
Can existing systems keep running?
Yes, subject to a technical review. We plan changes, backups and rollback paths to keep the transition controlled.
How are changes agreed?
Questions are collected in the project. Extensions get a defined scope and approval before additional implementation.
How is acceptance handled?
We agree concrete test cases and responsible people. Open issues are documented and release follows the agreed approval.
Is a handover session included?
Handover can include training, documentation and defined access rights. The scope and format are specified in the offer.
What happens after launch?
Support, updates and response times are agreed separately, making internal and KG-NET responsibilities clear.
How is personal data handled?
We limit access and review data flows, retention and processor arrangements. Legal approval remains with the controller and their advisers.
Can you work across borders?
We work with businesses in Austria, Germany, Switzerland and Serbia. Language, responsibilities and actual market needs are agreed together.
How are outcomes assessed?
We agree verifiable goals and identify the data source. Without connected measurement there are no invented metrics or guaranteed outcomes.
SELECTED WORK
Digital work. Real businesses.
Websites, stores and processes: explore selected projects and the services actually delivered.
ANKA
A webshop for ANKA connecting product discovery, user experience and payment integration.
TopMatic
A B2B system with a client portal, individual pricing and offer and invoice workflows.
GNT Logistics
A digital logistics presence that brings services, enquiries and connected information into one structure.
Rentacar ANKA
Vehicle presentation and reservations with pricing logic designed for car rental.
HBG Installationen
A structured service presentation for plumbing, heating and gas installations.
CleanPoint
A website with a booking system for cleaning and laundry services.
No references in this filter.
THE NEXT STEP
Your idea. A clear plan.
Tell us what you want to change. We will clarify goals, scope and the right approach, with an individual proposal.
KG-NET
Ilija Benjak
Völkermarkter Straße 280
9020 Klagenfurt am Wörthersee
Austria