DemoSynthetic data
See how scattered records become operational history.
The demo runs on a synthetic lab operation and shows AOI in two stages: first the structured events it pulls out of raw communications, then the operational memory it builds by connecting them over time.
01 · Messages, notes, logs, reports
Raw operational information
02 · Demo 1
Structured events
03 · Demo 2
Operational memory
04 · Next layer
Intelligence
One fictional organization, start to finish.
Both demos use the same synthetic dataset, so you can follow the same equipment, people, and problems from raw message to history. It happens to be a hospital lab. The mechanics are the same in a plant, a warehouse, or a hotel.
- Organization
- North Valley Medical Center, fictional
- Communications
- 226, all synthetic
- Sources
- Multiple source types
- Structured events
- 469, extracted by AOI
Entirely synthetic. No real organization, employee, patient, or operational data.
Demo 1
Structured Events
Raw communications become structured operational events. What to look for:
- Every event keeps its original source, so nothing is summarized away.
- Each event carries its date, entity, department, status, and severity.
- Statements are marked as stated, inferred, or uncertain, so you can see what the source actually says.
Huddle note · Day 4 · 07:05
Centrifuge 2 in the core lab is grinding again. Running everything on Centrifuge 1 for now.
Structured event
Illustrative- Entity
- Centrifuge 2
- Department
- Core lab
- Date
- Day 4 · 07:05
- Status
- Workaround in place
- Severity
- Moderate
- StatedGrinding noise on Centrifuge 2.
- StatedLoad moved to Centrifuge 1.
- InferredNot the first time: the note says “again”.
- UncertainWhat is causing the noise.
Source kept · Huddle note · Day 4 · 07:05
One event tells you what happened at one moment.
Operational memory asks what that moment belongs to.
- 01What else happened around the same entity?
- 02Did the activity come back?
- 03Was it mentioned through another source?
- 04What changed over time?
Entity history · Centrifuge 2
Illustrative- Recurring
- 5 events
- 4 sources
- Core lab
Episode 1 · Days 1 to 4
- D1 14:30Noise reported from Centrifuge 2.Email
- D2 09:15Belt tension adjusted.Maintenance log
- D4 07:05Grinding again. Load moved to Centrifuge 1.Huddle note
Original source · Huddle note
“Centrifuge 2 in the core lab is grinding again. Running everything on Centrifuge 1 for now.”
15 days later
Episode 2 · Day 19
- D19 06:50Same noise from Centrifuge 2 this morning.Message
- D19 10:20Work request opened.Maintenance log
Demo 2
Operational Memory
The same events, connected across time. What to look for:
- Entity histories: everything recorded about one instrument, area, or workflow, in order.
- Episodes: activity grouped into distinct stretches, so a return is visible as a return.
- Recurrence and relationships across sources: the same issue, mentioned in different places.
- Evidence drilldown: from a history, to an event, to the original record.
What the demos show, and what comes next.
The demos stop at operational memory on purpose. Analysis and reasoning have to sit on a history that can be checked, so we built that first. The later layers are in development and are not in the demos.
01 · In the demos
Extract
Pull events out of raw communications and records.
02 · In the demos
Structure
Give each event its fields: when, where, what, and how certain.
03 · In the demos
Connect
Link events that describe the same entity or activity.
04 · In the demos
Operational memory
Keep persistent histories, episodes, and recurrence.
05 · Next layer
Analyze
Look across histories for patterns worth a team’s attention.
06 · Next layer
Reason
Weigh what the evidence supports and what it does not.
07 · Next layer
Intelligence
Findings a team can act on, with the evidence attached.
Want to see what AOI could find in your operation?
We’re looking for a small number of operating teams to test AOI against real operational history.