The tradeoff of our mistake-resistant MCP architecture is that we don’t yet support querying all Factbird data at launch.
We are continuously improving our query support. The best way to find out what you can ask in that moment is to ask the agent what is supported, or simply try out the query that’s on your mind.
Contact support@factbird.com or your Customer Success manager if your desired query is not covered.
Supported queries (at time of writing):
Point it at a line, a factory, or your whole operation
Every query starts with telling Factbird which part of your operation you're asking about. You can name a specific production line, or point at a factory, plant, or area. Factbird automatically rolls up every line underneath it. Ask about "the whole company" and it'll cover everything you have.
Three ways to ask for data
A single number. Ask for one KPI over a time range - say, OEE for the last week. You get back one value.
A bar chart. Ask for one KPI broken down by group - by line, by day, by product, by stop cause, and so on. Useful for comparisons like "which lines had the most downtime this month."
A table. Ask for several KPIs at once, broken down the same way. Good for a fuller picture — for example, OEE, uptime, and scrap rate side by side, per line.
For bar charts and tables, you can break the numbers down by: line, directory (factory/plant/area), day, week, weekday, month, product, batch, shift name, or shift instance - or, for stop-related KPIs, by stop type, stop cause, or stop category.
Other charts are possible if you are willing to let your agent create the charts without Factbird’s help. This will give you more flexibility but more room for error.
What time ranges can you ask about?
- A specific date range - from one date to another, inclusive.
- The last shift.
- The last batch.
- Specific batches, if you know which ones you want.
What KPIs can you ask about?
Some of the most commonly used ones:
- Efficiency: OEE (three variants: 1, 2 and 3), TCU, uptime, yield rate
- Output: produced, good parts, scrap, scrap rate, planned amount
- Time: operating time, production time, manned time, value-adding time, longest run without a stop
- Speed: average produced per minute, speed while running/manned/producing, speed loss, cycle time
- Stops and downtime: downtime, number of stops, number of unregistered stops, micro-stop time, mean time between failures
Downtime and number of stops both count every stop, including brief "micro stops." If you only want to see certain kinds of stops ask your agent to filter for them.
Worth knowing
- Date ranges are inclusive on both ends.
- Always specify a time zone for date-based queries, so daily and weekly totals line up the way you'd expect. If you do not, your agent will most likely assume local time.
- You can ask about several lines or an entire facility in one go. You are not limited to one line at a time.
- Agents are clever enough to join different tables and results from intermediary tables into any format you’d like. Don’t feel limited to simple queries.