How to turn downtime causes into a knowledge base

Start with the shift, not the filing cabinet

A useful downtime knowledge base begins with a short, factual account of each stoppage. Record the machine, time, confirmed cause, corrective action and person who confirmed it. Add the relevant SOP, supplier note, part reference or maintenance instruction so the answer stays together.

Keep each entry written and searchable. That makes the record easier to find during the next shift than a note left in a chat message, spreadsheet cell or supervisor's notebook.

Use camera signals to trigger the right investigation

FactoryVision Factory Analytics uses one camera per machine to count finished work. It compares the count with the day's target and shows each station as on track, at risk or behind. This gives the team a clear reason to check a station while there is still time to act.

The count does not explain why a machine stopped. It shows where work is at risk and which waiting order depends on a station that is behind. Staff can then investigate and confirm the cause before adding it to the knowledge base.

Add visual checks and the reason for downtime

FactoryVision Visual QA checks parts as they are made. It can measure dimensions, inspect parts against a model of good work, check presence in marked zones, identify the right part for an order and watch process trends.

Use those checks alongside the production record. For example, note when a station moved behind, which part check was active, what was found and the confirmed reason for the stop. Do not treat a visual signal as proof of cause. Keep confirmed causes, observations and actions clearly separate.

See how to count finished work with one camera and how to check parts as they are made.

Store the written answer in one place

A complete downtime entry usually combines several kinds of reference: a confirmed cause, an SOP, a supplier note, a drawing or a previous fix. Put the written record in a knowledge base so the team can search by meaning rather than guess the exact wording used when the note was saved.

KnowMan saves articles and links with an optional note, searches them by meaning, answers questions with citations, keeps memories such as claims, decisions and playbooks, and provides an MCP server that lets AI agents search it. That makes it a separate place for written shop-floor knowledge, not a source of FactoryVision camera data.

Make every entry useful on the next shift

A searchable entry is only useful if another person can understand it. Use a consistent structure for the machine, symptom, confirmed cause, action and supporting reference. Include units, part names and supplier terms exactly as they appear in the work instructions.

Search before starting a repeat repair. A question such as “Why did the edge bander stop feeding?” can return the saved cause, cited answer and linked notes. If no confirmed cause was recorded, label the entry as an observation instead of filling the gap with a guess.

Build the knowledge base as work continues

Start with one machine and one recurring failure. Add only entries that a lead has checked. Over time, useful patterns may emerge, such as a cause that appears with the same part, supplier or adjustment.

A shop-floor knowledge base explains how written records can be organized for daily use. Keep the process small: capture the confirmed cause, save the supporting material and make sure the next shift can find it.

Turn the next confirmed cause into a reusable answer

Start a written downtime record, save the related SOP and supplier notes, and make the answer easy for the next shift to find. Request a FactoryVision pilot

Request a pilot