PHYGITALZBook Demo
AI AGENTS

Agentic AI Usecases

An agent does more than raise an alert. It reads your machine data and acts in the systems around the machine: maintenance tickets, the tool store, quality, purchasing and the schedule.

L1 Explains
Tells you what happened.
L2 Recommends
Suggests the action to take.
L3 Acts after approval
Does it in the software once a person approves.
L4 Acts on its own
Low-risk actions only, such as notifications, reports and reservations.
  1. Shift handover and morning brief agent

    Explains / Acts on its own

    Trigger The end of each shift.

    • Pulls the OEE losses, the top 3 downtime reasons for each cell, open breakdowns, SPC violations and tools due for change.
    • Writes a plain-language brief in English or Tamil, with an owner assigned to each action, and sends it by WhatsApp or email.
    • Tracks the next shift's open items.

    Result Every shift starts knowing what happened, what is open and who owns it.

  2. Downtime triage agent

    Acts after approvalAvailable now

    Trigger A machine stops for longer than a set time, with an alarm code.

    • Looks up the alarm in the Digital Machine Manual.
    • Checks past breakdowns on that machine and what fixed them.
    • Checks whether the likely spare part is in stock.
    • Creates a pre-filled breakdown ticket in Plant Maintenance.
    • Pages the right technician with a "probable cause and last fix" note.

    Result Shorter response and diagnosis time, because technicians stop starting from zero.

  3. Tool change readiness agent

    Acts after approval / Acts on its ownAvailable now

    Trigger A tool reaches about 85–90% of its life.

    • Checks tool store stock and reground availability.
    • Checks which tools the next job needs.
    • Reserves the tool and alerts the setter before the machine stops.
    • Drafts a purchase requisition if stock is below the minimum.

    Result Less waiting at tool change, and fewer premature changes.

  4. SPC drift response agent

    Recommends

    Trigger Quality 4.0 detects a trend or rule violation before parts go out of tolerance.

    • Correlates the drift with tool wear, spindle load, coolant or temperature, and recent tool changes.
    • Recommends a specific action to the quality engineer: an offset correction, a tool change or a machine check.
    • Flags the affected parts for 100% inspection through Traceability.

    Result A drift is caught and acted on before it becomes scrap. It recommends offsets; it never writes them to the machine.

  5. Predictive maintenance-to-action agent

    Acts after approval

    Trigger PredictM flags an abnormal condition such as vibration, temperature or current.

    • Checks spare stock and supplier lead time.
    • Finds a low-load window in the production schedule for planned maintenance.
    • Proposes the maintenance slot to the maintenance head and the production planner.
    • Drafts the purchase order if spares are short.

    Result A red alert becomes a scheduled job instead of an alert nobody acts on.

  6. Schedule recovery agent

    Acts after approval

    Trigger A breakdown on a bottleneck machine is expected to last longer than a set time.

    • Finds alternate machines with a compatible program, fixture and tool availability.
    • Works out revised completion dates for the affected orders.
    • Proposes a reschedule to the planner and, once approved, pushes it to the ERP.

    Result Affected orders get a revised plan that is ready for the planner to approve.

  7. Conversational plant analyst

    ExplainsAvailable now

    Trigger Anyone asks a question, by text or by voice.

    • Takes a question such as "Why was Cell 3's OEE 12% lower this week?".
    • Pulls the exact machine, cell, unit or plant data: OEE, downtime and production performance.
    • Answers in seconds, with a breakdown by loss category.

    Result The plant head gets an answer from live AIPLANT4.0 data without waiting for a report.

  8. Idle energy agent

    Acts on its ownAvailable now

    Trigger A machine sits idle beyond a set time while its hydraulics, coolant or chip conveyor are still running.

    • Alerts the operator or cell leader.
    • Logs the cost of the idle period.
    • Can flag an energy-per-part anomaly as an early sign of a machine problem.

    Result A measurable cost saving, from energy that was being used for nothing.