AI agents surge in Australia, but integration lagging

Salesforce research forecasts multi-agent adoption in Australia will rise by 73% over the next two years, as organisations expand the use of autonomous AI agents across teams and functions.

Its 2026 Connectivity Benchmark Report finds Australian organisations are already running an average of 11 AI agents, with 87% reporting that most or all teams and functions have adopted them.

Despite this uptake, many deployments remain fragmented. The report finds 50% of agents still operate in silos rather than as part of a multi-agent system, driving disconnected workflows, duplicated automations, and a higher risk of "shadow AI" being adopted without approval or oversight.

Integration pressure

IT leaders point to data integration as the central requirement for making agents useful across an organisation. In the survey, 98% agree success depends on seamless data integration across all systems, and 96% say IT architecture needs to become more API-driven for AI agents to succeed.

API use varies. The report finds 26% of teams already use APIs to speed up integration across systems, while 38% say they use APIs to connect and govern AI.

Alongside the push for wider adoption are concerns about complexity. According to the report, 96% of IT leaders worry agents will add more complexity than value when deployments run into integration challenges and agent silos.

Orchestration gap

The survey describes a growing orchestration and governance gap as organisations deploy agents more widely. It notes that infrastructure needs deeper integration for multi-agent environments to collaborate and use data securely across the enterprise.

It also points to application sprawl. The number of apps used by Australian organisations rose from 974 to 1,081 year over year, yet only 22% of applications are integrated.

Data barriers remain widespread. The report finds 98% of organisations face barriers to using data for AI use cases. It also finds 43% cite outdated IT architecture and infrastructure as a key blocker, with data silos and disconnected systems contributing.

Risk and skills

Respondents identified several hurdles slowing what the report calls "agentic transformation". Top challenges include a lack of internal expertise in AI and agent design; risk management and compliance or legal implications; limited trust in autonomous decision-making; and legacy infrastructure or system incompatibility.

Risk management, compliance, security, and legal implications emerged as a leading hurdle. The report also highlights cross-application data governance as a major integration challenge, with 51% of organisations citing it as a top issue.

Governance maturity appears uneven. The report estimates around one-fifth of APIs are ungoverned on average, and only 54% of organisations have a centralised governance framework with formal oversight for their agentic capabilities.

How agents are built

The report suggests many organisations run mixed environments, with agents coming from multiple sources. On average, agents are built using prebuilt SaaS agents (35%), embedded agents within enterprise platforms (33%), and custom-built in-house agents (31%).

Respondents also expressed interest in standards and protocols for managing and connecting agents. Interest levels include Agent Communication Protocol (45%), Universal Tool Calling Protocol (44%), Agent Network Protocol (41%), Agent-to-Agent Protocol (39%), and Model Context Protocol (39%).

Expectations for business impact are high. In the survey, 99% of IT leaders say agents have already improved employee experiences or expect them to do so, and 94% believe agents will free developers to focus on higher-value work.

Unified architecture

Salesforce positions API-driven approaches as a way to address fragmentation across systems and tools, describing APIs as a unifying layer that connects applications, data, and AI across an organisation.

"As agent adoption hits critical mass, AI agents are no longer experimental - they are becoming the primary driver of enterprise productivity," Salesforce said.

The report says the next phase of adoption in Australia will depend on stronger integration, more consistent governance, and architecture work that links agents with enterprise data and applications across the IT estate.