LLM Production Check
€750excl. VAT
One application, up to 30 test questions, a written report in 5 working days. Credited in full if a Deep Check follows within 30 days.
Four fixed-scope packages for teams with an LLM feature that is stuck or about to launch. Scope and price are agreed in writing before work starts. Automating a workflow instead? Fast deployment starts at €250 a day, and the estimator works out setup, hosting and processing for your setup.
LMXAI tests LLM applications before and after launch: answer quality, retrieval, tool calls, prompt injection, cost and EU AI Act evidence. The entry point is the LLM Production Check at €750.
€750excl. VAT
One application, up to 30 test questions, a written report in 5 working days. Credited in full if a Deep Check follows within 30 days.
€2,000excl. VAT
A larger test set, a prompt-injection suite, a model and cost comparison, and a reusable evaluation harness your team keeps.
€2,500excl. VAT
AI system inventory, risk classification, logging map, technical-file skeleton and gap list — engineering evidence for Regulation (EU) 2024/1689.
The service in detail€3,000excl. VAT
The Deep Check plus hands-on fixes and a before-and-after re-test. Longer work is quoted per project.
Fractional AI CTO and build projects are quoted per project. See the services.
AI workflow automation with n8n, Make or LangGraph, from €250 a day. Pick the systems, data sources and interfaces to connect: the estimator shows setup, hosting and processing separately and draws the system architecture as you go.
Yes. Scope and price are agreed in writing before work starts, and all prices exclude VAT. Work outside the agreed scope is quoted separately before it starts.
The LLM Production Check delivers its report in 5 working days. For the other packages the timeline is part of the written scope.
Access and data are used only for the agreed test. We can work under your NDA, and the scope records what we may access and when access ends.
You receive a report that shows what was tested and that it passed. That is evidence too: for your release decision and, where relevant, for EU AI Act documentation.
The Governance Pack covers inventory, risk classification, logging, the technical-file skeleton and a gap list. It is engineering evidence, not legal advice.
Yes. Send five questions your system must answer well. We run them and tell you what we see, free of charge.