Reports are rebuilt from exports
Someone pulls data out of three systems every Monday, so the numbers are a week old and only as right as the spreadsheet.
AI & Analytics
Reads the business as it is. Leaves the decision to you.
Analytics and AI read the same live records every other app writes. A report is a view over those records rather than an export, and the AI watches what you ask it to, explains what it noticed, and changes nothing without a person’s approval.
The problem
Where the insight arrives too late.
Most businesses find out what went wrong. Fewer find out while there is still time to do something about it.
Someone has to notice
A late order is spotted when the customer rings, because nothing was watching the delivery date for you.
Dashboards disagree with the ledger
A chart built on a copy of the data drifts from the books, and the meeting is spent arguing about which is right.
Automation is a script nobody owns
The rule that emails a customer was written by someone who has left. Nobody knows what it sends, or when.
Capabilities
What the AI & Analytics app covers.
Five groups over the same Core OS records. Two of them are in early access and say so.
WatcherEarly access
A watch on any record, raised when something needs attention.
- Deadlines
- Value limits
- Inactivity
- Observations
- Resolve
Agent workflowsEarly access
A process drawn on a canvas, with a person’s sign-off in it.
- Triggers
- Conditions
- Agent steps
- Human approval
- Versions
Analytics
One reporting surface over every app.
- Dashboards
- Reports
- Branch views
- Trends
- Drill-down
Forecasting
Demand read from live movement, not a monthly export.
- Inventory forecasting
- Reorder points
- Replenishment suggestions
- Slow movers
Automation
Rules that act on business events without anyone watching for them.
- Business events
- Schedules
- Notifications
- Assignments
- Audit trail
Workflow
One observation, six stages.
The watch, what it noticed, what the agent found and who signed off are kept together on the record they were about.
An observation moves through six stages: a watch is set, a rule is met, an observation is raised, an agent investigates, a person approves and the observation is resolved.
-
01 Watch
SO-2026-0148
Set
-
02 Rule
SO-2026-0148
Met
-
03 Observation
SO-2026-0148
Raised
-
04 Agent
SO-2026-0148
Investigated
-
05 Approval
SO-2026-0148
Signed off
-
06 Resolved
SO-2026-0148
Closed
Use cases
Where it earns its place.
A sales manager with too many open orders
A watch on every order due this week, so the manager hears about the three that are slipping instead of checking all forty.
A buyer replenishing hundreds of items
Reorder points that follow what is actually being consumed, so the buying list reflects this month rather than last year.
An owner who wants one honest view
One dashboard over sales, stock and receivables, read from the books themselves, so the weekly meeting starts from the same numbers.
Works with
AI & Analytics on its own is useful. Connected, it is the point.
These read and write the same Core OS records, so a handover between them passes a reference rather than a retyped copy.
Sales
Quotations, sales orders, pricing and customer terms.
Inventory
Real-time stock across warehouses, batches and serial numbers.
Purchase
Requests, purchase orders, supplier pricing and receipts.
Finance
Invoicing, receivables, payables and one double-entry ledger.
FAQ
Questions about AI & Analytics.
Does the AI change records on its own?
No. A watch observes and notifies; it never changes a business record. In an agent workflow, anything that would go out waits for the approval step you put in, and who approved it is recorded.
What does early access mean here?
Watcher and agent workflows are in use today and still being finished. Some ways of starting them — on a schedule, or the moment a record changes — are not switched on everywhere yet, so we will tell you exactly what is live for your site before you rely on it.
Where does the data come from?
From the same Core OS records the other apps write: orders, stock movements, invoices and payments. There is no export, no nightly copy and no separate warehouse to keep in step.
Do we need a data team to use it?
No. Reports and dashboards are built on records the apps already keep, with filters for branch, period and customer. Watches and workflows are set up on screen, without code.
Have a different question? Talk to us.
See it watch one of your own orders.
Bring an order that went wrong last month. We will show you the watch that would have caught it, and when.