Sentinel
The system that remembers so your team doesn't have to.
Sentinel turns fragmented operational evidence — SAP exports, field logs, engineer notes, machine events — into structured, auditable, human-approved case states. The record your institution can trust, governed end to end.
The machine proposes. The human authorises. The system remembers. The institution improves.
Governance first
Governed memory,
not just software.
Most tools store data. Sentinel governs it. Every entry in the operational record begins as a proposal: the machine reads the evidence and suggests a memory write.
A human authorises it. A named person approves, edits, or rejects each proposed fact before it becomes part of the record. The machine never writes memory on its own.
Then the institution remembers it. Approved case states are permanent, searchable, and attributable. When staff move on or contractors rotate out, the knowledge stays where it belongs.
Auditability is not a feature you add later. It is the foundation everything else sits on.
How it works
From raw evidence to compounding memory.
Four phases turn whatever you already produce into a governed record that gets more valuable with every authorised decision.
Intake
Raw evidence arrives as it already exists — SAP EAM/PM exports, CMMS records, field logs, engineer notes, machine events. Nothing is reformatted by hand.
Seed
Structuring agents read the raw evidence and extract candidate facts: which asset, which fault, what was done, what it cost. Each is proposed, never assumed.
Live
A human review queue puts every proposed memory in front of a person. Approve, edit, or reject. Nothing enters the record without authorisation, and every decision is attributed.
Compound
Approved memory powers daily briefs, recurring-fault detection, and asset health scores. Each authorised case state makes the next decision faster and better evidenced.
Proof
What governed memory does to the numbers.
Figures below are pilot averages and illustrative. Your diagnostic establishes the baseline we measure against on your own data.
73%
reduction in unplanned downtime
Pilot average
89%
of issues documented and attributed
Pilot average
45%
lower repeat-contractor cost
Pilot average
4 weeks
from kick-off to live on your data
Standard pilot
Built to fit your stack
It reads what you already run.
Sentinel sits over your existing systems. No migration project, no second source of truth — your evidence flows in where it lives.
SAP EAM / PM
Notifications, orders, and confirmations flow in as structured evidence — no rip-and-replace.
CMMS
Work history and asset registers from your maintenance system seed the case record.
SCADA / IoT
Machine events and sensor signals attach to the assets they actually concern.
Field logs
Engineer notes, photos, and contractor findings captured in the field, not lost to inboxes.
Pilot structure
Start with a four-week pilot.
We stand Sentinel up on your live evidence in four weeks: intake connected, structuring agents seeded, your review queue running. You see governed memory on your own assets before you commit to anything longer.
Then from
£8k–15k
per month
How the engagement runs
- 01
Diagnostic
ApexIndex maps where memory leaks and sets the baseline. From £2,500.
- 02
Four-week pilot
Sentinel goes live on your data. Intake, seeding, review queue.
- 03
Monthly service
Governed memory in production at £8k–15k per month.
- 04
Annual contract
Proven value rolls into a committed annual partnership.
The diagnostic routes the pilot. It tells us exactly what to connect first.
See Sentinel on your own data.
Bring a sample of your operational evidence. We will show you what governed memory looks like on the assets you actually run.