This year’s Earth Day (22 April), the conversations pivot from carbon accounting to carbon action. While APAC CIOs have embedded sustainability dashboards, the rise of autonomous agents threatens to undo this progress. 

In 2026, an uncontrolled "agent sprawl" could exponentially increase compute, data storage, and energy use—directly conflicting with Net Zero pledges. True sustainability isn’t just about reporting emissions; it’s about embedding green governance into every autonomous decision. 

As agents become "digital coworkers," CIOs must treat energy efficiency and waste reduction as non-negotiable compliance metrics, ensuring AI acceleration doesn't come at the planet’s expense.

With us to understand what Earth Day means in the context of the exploding AI agent sprawl is Mr Liher Urbizu, present and MD of SAP Southeast Asia.

Questions covered:

1.       Give us the agentic sprawl in Southeast Asia in 2026.

2.       How do you embed "carbon-aware" policies directly into agent workflows? This should force autonomous agents to defer non-urgent batch processing to times of renewable energy availability. (treat carbon data like financial data)

3.       With Earth Day commitments tightening, what technical controls are required to mandate energy consumption caps per agent, treating efficiency as a governance rule rather than a post-execution report?

4.       To ensure agents don't inadvertently increase waste, how do you establish trusted data lineage for Scope 3 emissions, enabling an agent to verify a supplier's carbon intensity before autonomously placing an order?

5.       Given that poor data quality leads to redundant processing, what data governance rules are necessary to prevent agents from repeatedly querying or transforming the same inefficient datasets, wasting energy?

6.       How do you build a "sustainability audit trail" for every autonomous decision, allowing CIOs to trace a specific agent's action back to its energy cost and carbon footprint for regulatory reporting?

7.       As we manage agents like digital coworkers, what "retirement criteria" ensure that low-value, high-frequency agents are automatically decommissioned to prevent long-term energy leakage? (leanIX)

8.       To avoid "shadow agent" sprawl doubling your infrastructure emissions undetected, what discovery tools can catalog every autonomous agent and calculate its real-time energy consumption against your Net Zero milestones?

9.       With stakes higher than Shadow IT, how do you differentiate between essential agents that optimize sustainability (e.g., logistics routing) versus "rogue" agents that create unnecessary digital waste and technical carbon debt?

10.   Where is the starting point for my organisation to move towards a more sustainable IT operation?