AI Core Hub
✨ Setup

Set up AI Core Hub in two quick steps.

First install the Hub into your SAP system, then connect it to your AI model. Both are fast — you can be live the same day.

1 Install into SAP ≈ 2 h → 2 Connect the model 1 h – days
Step 1 · Installation

Install it into your SAP system.

A set of packages and objects loads straight into your existing ABAP stack — via abapGit or a transport request. No new servers, no migration.

AI Core Hub PACKAGES & OBJECTS packages · classes reports · skills tables · SICF nodes abapGit / transport request YOUR SAP SYSTEM ✓ installed ⏱ ≈ 2 hours — then done no new servers · no BTP
The Hub's packages and objects load straight into your ABAP stack — via abapGit or a transport request. About two hours, and it's installed.
✅
One requirement: an ABAP stack on SAP_BASIS 7.40 (SAP NetWeaver AS ABAP 7.40) or higher — ECC or on-prem S/4HANA both qualify.
Step 2 · Model integration

Connect it to the brain.

AI Core Hub calls out to the AI model you choose — every call masked first. There are three ways to make that connection.

YOUR SAP SYSTEM AI Core Hub reads your SAP screen your data stays inside 🔒 masked every call 🧠 Your existing model OPTION 01 · easiest 🧠 A model we set up with you OPTION 02 · guided 🧠 Our optimised gateway OPTION 03 · coming soon
Every call is masked as it leaves SAP — then routed to whichever brain you choose. Three ways to connect; pick the one that fits where you are today.
⏱
Connecting the model: about an hour for one you already run — up to a couple of days if we set a new one up, depending on complexity.
The three ways to connect
Option 01◆ Easiest

Onboard the model you already have

You already run an AI model in the company — we simply connect the Hub to it.

  • Point the Hub at your existing endpoint — Azure OpenAI, Bedrock, Vertex, or any compatible gateway.
  • Your model, subscription and keys stay exactly as they are.
  • Business values are masked on the wire before anything leaves SAP.
  • Live in hours — lowest lift, lowest cost.
🔑
What we need from you: the endpoint URL, the model name, and an API key (or nothing, if your gateway is keyless).
Option 02◆ Guided setup

We help you establish & connect one

No brain yet? We help you set up an account with Azure (or another provider), configure it, and connect it.

  • A ready-to-run blueprint provisions the model service in your Azure / AWS / GCP tenant.
  • A step-by-step runbook: model choice, region, private networking, keys.
  • We do the Hub-side wiring and validate the first live conversation with you.
  • You end up owning the subscription, keys and data boundary.
🧭
What we need from you: a cloud subscription and an approver. Typical time-to-value: days.
COMING SOON
Option 03◆ In development · on our roadmap

We become your model provider

Let us run the brain for you — one optimised, resold gateway to the best models. This option is in development.

  • Zero infrastructure for you — we operate the endpoint, you consume it through the Hub.
  • Optimises your usage — the Hub routes each request to the most cost-effective, best-fit model for the job, so you never overpay for a heavy model on a simple question.
  • We act as your reselling partner: one bill, one relationship, better rates through aggregation.
  • Routed only to compliant, EU / German-hosted providers; still masked on the wire.
🧭
Status: on our roadmap — coming soon. Tell us if you'd like early access when it lands.