Fast Tech, Slow Org: The AI Execution Gap | Salesforce SVP & CMO ANZ Leandro Perez

Subscribe if your leadership team is investing in AI but still cannot clearly explain what work has changed.

AI progress is easy to perform.

Pilots launch. Platforms get bought. Agents get tested. Boards receive updates.

Value is harder.

It shows up when workflows change, ownership is clear, metrics move, customers get a better outcome, and the organisation knows what the agent is allowed to do.

That is the AI execution gap.

In this episode of Applied AI Australia, powered by Acquire Intelligence, Ramon Rodriguez speaks with Leandro Perez, SVP & CMO ANZ at Salesforce, about what has to change before AI capability becomes business value.

The conversation goes past tools and into the operating reality: agents need owners, metrics, supervision, escalation paths, business context, clean data and better work design.

Leandro also raises one of the sharper leadership shifts: this may be the last generation of managers who only manage people. As agents enter workflows, leaders will need to manage both people and AI systems.

The core message:

The tool is not the transformation. The proof is whether the work changed.

What you’ll learn:

• Why AI pilots, tools and training do not automatically create business value
• What changes when AI moves from answering questions to acting inside workflows
• Why agents need owners, metrics, supervision and escalation paths
• How poor data and weak context limit AI performance
• Why customer service, sales and marketing metrics need to change
• What leaders should ask before scaling AI agents

Key stats and examples:

• 76% of Australian service leaders are looking at or using AI
• 56% of employees are using personal AI tools without disclosing it
• Salesforce says 84% of cases are now handled autonomously in one support use case
• More than 2.2 million cases have been deflected through Salesforce’s support agent
• Salesforce has more than 300 agents operating internally
• One prospecting agent had to be slowed down after following up outside normal business expectations

48-hour action: Pick one AI agent, pilot or workflow and ask:

1. What work has changed?
2. Who owns the outcome?
3. What metric proves value?
4. What authority does the agent have?
5. Where does a human step in?
6. What outcome has improved?
7. What needs to change before this scales?

If you cannot answer those questions, you may be scaling AI activity, not AI value.

“Treat agents like an employee. You don’t just hire someone and leave them in the corner.” - Leandro Perez

Chapters:

• Fast tech, slow org
• AI activity versus AI progress
• Why AI pilots fail to become value
• Why personal AI does not equal enterprise AI
• Salesforce’s internal agent lessons
• Why business experts need to manage agents
• Data, privacy and guardrails
• Who owns an agent when it goes wrong?
• How to set success metrics for agents
• Why AI changes old performance metrics
• What leaders should do next

About Applied AI Australia:

Applied AI Australia is powered by Acquire Intelligence. We help Australian companies turn AI into revenue, margin, time back, and better operating discipline.

Need immediate execution support?

From 10 December 2026, APP 1.7 requires organisations using automated systems to make, or substantially influence, decisions about individuals to disclose that fact in their privacy policy. Serious or repeated privacy interferences can carry penalties up to $50 million.

If your AI agents, scoring tools, triage systems or automated workflows touch customer decisions, you need a decision map before you can know what must be disclosed.

Acquire Intelligence runs six-week, governance-led sprints to build compliant decision maps from scratch before enforcement arrives.

Websites:

www.appliedaiaustralia.com.auwww.acquireintelligence.ai
Ramon:linkedin.com/in/ramonrod

Guest:

Leandro PerezSVP & CMO ANZ, Salesforcelinkedin.com/in/leandro-perez