Local processing
A local model processes documents. Permissions, storage and controlled service access are part of the design.
LOCAL AI FOR MANUFACTURING
Local AI for enquiries, technical documents and company knowledge. We deliver the hardware, connect it to a specific business process and take care of ongoing operation.
Launch virtual tour360°
01 / EVERYDAY WORK
We start with one process where your team spends time copying, finding or checking information.
Choose a task to see an example output.
Prepares a list of requirements from an email and its attachments. Flags missing or conflicting information for an engineer to check.
Source: fictional enquiry, attachment 01
Less copying. Ambiguities are visible before the quote is prepared.
Document processing scope and accuracy are tested on your cases before production use.02 / ON YOUR PREMISES
A capable server, an AI model and connections to your systems. Designed to process the agreed tasks within your network.
A local model processes documents. Permissions, storage and controlled service access are part of the design.
Compute, storage and connections depend on your documents, concurrent tasks and required response time.
You purchase the AI Box. Process implementation and ongoing care are specified in your proposal.
03 / MANAGED BY OPENROKK
An AI model is software that reads instructions, works with text and prepares answers. Openrokk selects the right version, configures its behaviour and tests it on the tasks it needs to handle.
Selected, tested and managed by us.
Connected documents and company systems.
Configured rules, tools and approvals.
04 / OPENROKK STAYS INVOLVED
Openrokk brings AI into company operations. With AI Box, we bring the hardware, model and business process together.
We review the current workflow with your people. We agree what AI should prepare and how we will assess its value.
We connect sources, configure permissions and test outputs. Your people try the solution on their own work.
We train users, set approval steps and a fallback procedure. Production use starts after agreed acceptance tests.
Under the agreed service plan, we monitor operation, resolve faults and test model updates before deployment.
Acceptance tests and responsibility for fixing defects are agreed in advance. Warranty, support scope and response commitments are specified in your contract.
Purchase
Hardware, a local model and installation. Configuration is sized for the first process.
Implementation
System connections, permissions, testing and training. We verify the result together at handover.
Ongoing care
Fault resolution, monitoring and tested updates under your agreed service plan.
Your proposal separates purchase, implementation and recurring care. Additional processes and major extensions are agreed separately.
05 / YOUR QUESTIONS
AI Box hardware, a configured runtime and a model. Together we define a specific process implementation and ongoing care. The proposal separates the initial delivery from recurring services.
We build on open models. We select, configure and manage the delivered version. We are preparing a fine-tuned version for manufacturing documents; deployment depends on test results. We disclose the origin and version of the model used.
For a local process, document processing, the model and storage are designed to run within your network. We also agree updates, backups and service access with your IT team. Any external connection is explicitly agreed in advance.
People approve critical information and final outputs. We prepare a manual fallback and agree support during implementation. Model updates are tested first, with rollback to the previous verified version if needed.
This depends on the hardware, system connections and scope of the first process. We propose the next step after an initial conversation. Pricing, delivery dates and ongoing care are set out in your proposal.
START WITH YOUR OPERATIONS
Bring one recurring task to the first conversation. Together we will find out whether AI Box is a good fit.
Download overview (PDF, Czech)