1300 166 138 Tell us what you need
News / AI Agents

Databricks puts $400 million behind SAP’s autonomous agents

By Monika Tantau • Published 6 months ago
Databricks puts $400 million behind SAP's autonomous agents

Key takeaways

• SAP unveiled an Autonomous Suite of 50+ Joule assistants orchestrating 200+ agents across finance, supply chain and HR
• Databricks committed $400 million to expand its A/NZ presence as a strategic partner on the Business Data Cloud
• KPMG has enabled 3,500 consultants with SAP Joule as a Global Strategic Service Partner
• Joule Studio 2.0 is free until year-end and the AI Agent Hub launches in Q3 at no extra cost

Most Read

SAP used its Sapphire 2026 conference this week to set out an ‘autonomous enterprise’ vision, where AI agents move from assisting employees to running entire business processes end-to-end. The announcement reshapes how Australian and New Zealand customers should think about their next ERP investment cycle, and it pulled an unusually direct response from local partners.

Databricks, named as a strategic partner on the new Business Data Cloud, committed $400 million to expand its A/NZ presence alongside the launch, according to ARN. The signal is plain: enterprise AI’s next phase is being decided in the implementation layer, not the model layer.

What the autonomous suite actually does

At the centre of the announcement is the SAP Autonomous Suite, which will deploy more than 50 domain-specific Joule assistants that orchestrate over 200 specialised agents across finance, supply chain, procurement, HR and customer engagement. SAP framed the shift in simple terms: Joule no longer just suggests, it executes. Users describe an outcome through Joule Work and the platform sequences the steps. Developers get Joule Studio 2.0 free through year-end to build their own agents, and an AI Agent Hub arrives in the third quarter at no additional charge.

The change in scope is meaningful. Where last year’s Joule was a chat-style helper for individual tasks, the new orchestration model lets a single outcome cascade across finance, procurement and HR without manual handoffs.

SAP cited live deployments at KPMG, Ericsson, Bayer, Novartis, H&M and JPMorganChase, alongside agent-led migration tooling that the company says cuts migration effort by roughly 35 per cent. None of those names are Australian, which is part of the read here: the first wave of customers is global enterprise, with A/NZ adoption likely to follow through local partners.

The platform layer underneath

The harder bet sits underneath the agents. SAP’s new Business AI Platform unifies its Business Technology Platform, Business Data Cloud and AI capabilities, and introduces what SAP calls a ‘company memory’: a context graph holding policies, procedures and approval rules. That memory is the governance answer to the loudest concern about agentic AI in the enterprise: agents that act on stale or inconsistent data.

Anthropic and Nvidia were named as model and protocol contributors, and SAP cited SOX audit compatibility, full logging and traceability across the stack. The pitch is that this is the first agentic platform built for finance and audit teams, not just developers. Whether it actually clears those bars is the question every CFO and CIO will be asking through the rest of the year.

Where Australian partners see opportunity

For Australian and New Zealand customers, the implementation reality matters more than the keynote. DXC Technology’s A/NZ SAP practice lead called the vision sound but said success depends on staged adoption. New Zealand partner Bonfire Consulting welcomed SAP’s openness to third-party agents, which reduces lock-in risk and lets customers integrate non-SAP tooling without rebuilding their stack.

KPMG, which has now enabled 3,500 consultants with SAP Joule and sits as a Global Strategic Service Partner, used the launch to argue that SAP’s approach finally puts governance and orchestration into a market that had been operating largely without it. Syniti and other data partners were blunter: data quality remains non-negotiable for any of this to work, and most A/NZ enterprises still have meaningful data hygiene gaps before agentic execution is realistic.

Why Australian businesses should pay attention

The most concrete A/NZ signal came from Databricks, which committed $400 million to expand its regional presence in lockstep with the SAP partnership. The money lands in a market where partner contribution is already doing more of the heavy lifting. AWS recently disclosed its A/NZ partner contribution moved from 47 per cent to 68 per cent.

For Australian operators, the implication is that the next twelve months will be defined less by which model a vendor uses, and more by which partners can wire the platform into compliant, governed workflows. That is consistent with what Australian operators see in production work involving AI consulting and process redesign: the value sits in implementation discipline, not the demo.

Compliance teams will want to look closely at SAP’s claims around SOX auditor compatibility, traceability and logging. Those features are what differentiate an autonomous agent that a CFO can sign off on from one that quietly creates exposure. For ASX-listed companies and large privately held operators in Australia, the auditability bar is not optional. That is the question that will determine whether the agent hub in Q3 actually moves into production or stays in a sandbox until the auditors are comfortable.

Why Australian businesses should pay attention

A new bar for enterprise automation

Sapphire 2026 made one thing clear: vendors are no longer competing on assistants. They are competing on whether their agents can complete the work and stand up to an audit while doing it. For Australian businesses still piloting GenAI, that raises a different planning question. Which processes are actually safe to delegate, and which need the human in the loop because the data underneath is messy?

The smart move is not to wait until 2027 to start mapping that distinction; the mapping itself is most of the value. For document-heavy workflows, document automation is the obvious starting point, because the data quality work that agentic execution requires is forced upfront.

Scroll to Top