Applied AI laboratory / Ostrava
We help small and medium companies find out whether AI can handle their specific task.
Our own technical facility in Ostrava.
Evidence for your decision
The output is evidence for a decision: what the model handles, what a deployment needs and where people stay in charge.
See the example protocol01 / Ways to work together
We start from the problem you are solving. Scope, availability and the delivered output are agreed individually.
Can AI handle a task our people do today?
We help define the problem, the expected benefit and the success criteria. Then we pick the assumptions worth testing before further investment.
Service detailsWhich solution has the quality and handles our workload?
We compare selected models and configurations on your specific task. We look at correctness, latency, memory and behaviour under load.
Service detailsShould it run on-premise, in the cloud, or both?
We assess the technical options, data requirements and the fit with your systems. A prototype or integration work can follow by agreement.
Service details02 / The engagement
Before testing we agree how to recognise a useful result. Finding where a solution falls short matters just as much.
What you solve today, how you do it and which decision needs evidence. You do not need a model picked in advance.
Together we define the goal, the inputs needed, the evaluation method and the delivered output. We confirm price, timing and data handling.
We compare the agreed options and record both successful and failed cases. The Lab team runs the test, or you use its infrastructure by agreement.
We explain the findings, the limits and the options. The outcome can be a recommendation to deploy, to test further, or to stop.
03 / Example output
A test answers one specific question. The output covers the measurement conditions, the limits we found and a recommended next step.
The outline below explains the approach on three briefs. These are not results of real projects or measured data.
Question to answer
Dozens of e-mails. The same fault described in different ways. Let us split one large brief into questions that can be verified.
E-mail → fault type → the right technician
A real report adds, within the agreed scope, the models and configurations used, the evaluation method and the results including failed cases.
04 / Facility and people
We bring together developers, AI specialists and people from companies. Technical knowledge is verified on real briefs and our own infrastructure in Ostrava.
A server room, a server cluster and an NVIDIA DGX Spark unit support model comparisons, load testing and checks of local deployment.
The exact configuration, capacity and access conditions are confirmed for each brief.
The facility and people in detailMembers bring topics, knowledge of their fields and experience from practice. Together we develop tools, share findings and look for questions worth testing.
First members join by invitation or after an individual review. Participation and access to the facility depend on agreement and capacity.
The facility and people in detail
Founding member
AI agents · DevOps · MLOps
Combines experience in development, cloud infrastructure and systems operations with artificial intelligence. Works on AI agents and putting them into practice.
Profile of the founding memberMeetups / Events
Events we take part in. A chance to talk in person about AI, technology and experience from practice.
Lukáš Orčík / Speaker
Founding member of the Lab Lukáš Orčík gives the talk “Czech applications talk to AI: MCP connectors in practice” (in Czech). It covers connecting Czech applications to AI agents and practical use of MCP.
Černá stage · 15:45–15:57 · workshop “From transport to microgreens” 12:30–13:00
Details on the organizer's website05 / Practical questions
Basic orientation for a first conversation. The details always depend on the specific brief.
No. We can start from a specific problem and a description of the current process. The choice of model and test method follows from the required quality, the available data and the constraints.
Briefly describe the task, who performs it today and what should improve. An example input with a correct result helps. Share sensitive data only after we agree how it is processed.
By the scope of the brief, the tests needed and available capacity. Before we start we agree the output, the price and the deadline. A short consultation and prototype development differ in scope.
Yes, local deployment is one of the tested options. We assess which model and hardware fit the task and what administration and operation will require.
For selected equipment, by individual agreement. The purpose, available capacity, duration and access conditions have to be confirmed in advance.
A prototype verifies an agreed assumption or function. Integration, security, monitoring and follow-up support are defined separately.
06 / Contact
For information about working together, the community or partnerships, contact the operator. A short description of your problem is enough.
kontakt@orcik.techAvailability and scope are confirmed individually.
Operator of Ostravský Lab
Orcik Tech s.r.o.Company ID: 08763941
VAT ID: CZ08763941
Jakoubka ze Stříbra 781/44
Nové Sady, 779 00 Olomouc
Registered office · the Lab's technical facility is in Ostrava
Operator's website