Skip to content
aumet.ai
How it works

From your data to an approved decision.

The layer sees, understands and decides. People approve. Your systems execute. Here is what happens at each step, what you connect, and how the first month runs.

01 · See

The five stages

Connect

Aumet connects read-only to the systems you already run, or starts from a simple template. Data quality is measured and shown, never hidden.

  • Stock by site, lot and expiry
  • Sales and consumption history
  • Purchases, orders and deliveries
  • Sites, transfer lanes and storage such as cold chain
02 · Understand

One picture

The layer models supply, demand, expiry and clinical criticality as one network. Statistical models and an optimizer produce every number; a language model never does.

  • Demand per product, site and period, with seasonality
  • Stock at risk of expiring unused, valued at cost
  • Supplier and delivery reliability
  • Zero sales during a stockout is never read as no demand
03 · Decide

Ranked and explained

Agents turn each risk into a feasible plan and rank it by clinical criticality, days left to act and value. Governance checks every plan against your policy with fixed rules.

  • Every plan shows its passed checks
  • Options it rejected, and why
  • Who must approve it, by role and value
  • Unknown policy means blocked
04 · Approve

People decide

Each decision goes to the person who owns it. They approve, adjust or reject it. Nothing moves before that.

  • Approval matrix by role and value
  • Edit quantities before approving
  • Every action logged with its owner and time
05 · Execute

Act and learn

Approved decisions are written back to your systems as transfers, requests or orders, and followed until they are received. Finance verifies the value.

  • Write-back in your ERP’s own format
  • Each saving counted once, from exposure to verified
  • Outcomes improve the next forecast
Deployment

Runs where your data is allowed to be.

Data and models stay in Aumet’s infrastructure in the region or in yours. No outside AI service is used.

Aumet cloud, in region

The fastest start. Hosted in the region, isolated per organization, managed by Aumet.

Your infrastructure

For national programs and hospitals that must keep data on their own servers or national cloud.

Offline action bundles

Where systems cannot accept write-back yet, approved decisions are exported in the format your team imports.

The first month

From connection to first decision.

Week 1Connect or upload. Data status report.
Week 2Rules configured with your administrator. Roles invited.
Week 3First ranked decisions. Approvals begin.
OngoingWrite-back, action tracking and verified results.
Questions

About how it works

What data does Aumet need to start?

Sites, products, stock by lot and expiry, and sales or consumption history. Purchases and deliveries improve the results. It can start from a read-only ERP connection or a CSV template.

How long until the first decisions?

Typically about three weeks: one week to connect data, one to configure rules and roles, and ranked decisions from the third week.

Which systems does Aumet connect to?

ERPs such as Odoo and SAP, hospital information systems, pharmacy point-of-sale systems, and Aumet’s own Pulse, Marketplace and ERP, connected per deployment. Where no direct connection exists yet, a CSV template works.

What if our data is incomplete?

The layer says so. Data status is shown per site and product, missing costs or history are flagged, and decisions that depend on missing data are held back rather than guessed.

Where does Aumet run?

In Aumet-controlled infrastructure or on your own, as set in your contract. Your data is not sent to an outside AI service.

See the pipeline run on your data.

Bring a stock export. In the first session you will see the value at risk, the first decisions, and the reasons behind each one.