01
Live data, not a warehouse dump
The assistant queries the same real-time feeds that power your OEE dashboards. No stale extracts, no next-day batches, no ETL pipelines to maintain.
02/08 · AI ASSISTANT
Ask your factory a question. Get an answer in seconds.
xynLog ships with a built-in AI assistant that speaks to your live CNC data. Ask it why a machine stopped, which operator ran worst yesterday, or how today's OEE compares to last week — and get answers without opening a BI tool.
Works with the LLM you trust:
OpenAI GPT-4oAnthropic ClaudeAzure OpenAILocal OllamaYour own model
A factory chatbot that actually knows your machines
02
Your language, literally
Ask in English, Czech, Slovak or German. Responses mirror the question's language, with machine names, shift codes and the terminology your shop already uses.
03
Natural-language OEE
'Which machine had the worst performance today?' 'Why was shift 2 slower than shift 1?' 'Show me downtime over 20 minutes.' — No query language, no training.
04
Choice of LLM
OpenAI GPT-4 class, Anthropic Claude, Azure OpenAI, or local Ollama. You pick per deployment based on cost, latency and data residency.
05
On-prem option
Run the assistant entirely inside your network with local models. Your process data never leaves your plant — ideal for regulated industries.
06
Explained answers
Every answer links back to the underlying chart and data. Click through from a number in the chat to the exact dashboard the assistant is reading.
Who uses it
Plant managers
Daily questions like 'What hurt OEE today?' or 'Which operator hit target?' answered in seconds — without opening a BI tool or waiting on analytics.
Shift supervisors
Downtime reasons, runtime comparisons and alerts — ask and act without leaving the shop floor. Works on tablet and phone.
IT and compliance
On-prem option, full audit trail of every question and answer, GDPR-friendly EU hosting when cloud is acceptable.
Frequently asked questions
Out of the box: OpenAI GPT-4o, Anthropic Claude Sonnet, Azure OpenAI deployments, and on-prem Ollama with Llama 3 or similar. Other models can be wired in on request.