Agentic AI Is Here. Are Indonesian Enterprises Ready to Let AI Act?

For the past few years, most companies have experienced artificial intelligence as something they ask.

Write this. Summarise that. Analyse these numbers.

Agentic AI changes that relationship.

An AI agent can potentially plan a task, interact with different applications, make decisions and carry out actions without waiting for a person at every step. Gartner expects 40% of enterprise applications to include task-specific AI agents by the end of 2026, up from less than 5% in 2025. Read Gartner’s forecast

That was the question at the centre of Techpresso: “Agentic AI, Are We Ready?”, held by Medcom.id on 10 September 2026.

The discussion brought together Nezar Patria, Vice Minister of Communication and Digital Affairs, Hammam Riza, AI and Digital Transformation Expert at the National Research and Innovation Agency (BRIN) and Patrick Dannacher, President Director of PT ITSEC Asia Tbk (IDX: CYBR).

Medcom.id reported that the session examined Indonesia’s readiness from the perspectives of regulation, infrastructure, research and cybersecurity. Read the Medcom.id coverage

The Question Changes Once AI Can Act

During the session, Patrick used a simple comparison: if an employee is given access to a company system, the organisation normally decides what that employee can see, what they can change and what they are authorised to approve.

An AI agent should be treated with the same discipline.

An agent connected to email is one thing. An agent that can access an ERP system, customer database or financial workflow is another.

The practical questions become:

• What systems can the agent access?
• What actions is it allowed to execute?
• Does that access remain active after the task is complete?
• Can every decision and action be traced afterward?
• Who is accountable if the agent gets it wrong?

Patrick stressed access control during the Techpresso discussion, arguing that organisations need to define the boundaries given to AI agents in much the same way they manage access for employees. Watch Patrick’s Metro TV segment

The risk is no longer limited to an AI producing an incorrect answer. A wrong answer connected to sufficient authority can become a wrong business action.

Adoption Is Moving Faster Than Enterprise Discipline

Interest in Agentic AI is already high. Gartner says only 17% of organisations have deployed AI agents so far, while more than 60% expect to deploy them within the next two years. Read Gartner’s Hype Cycle for Agentic AI

There is another number worth paying attention to. Gartner predicts more than 40% of Agentic AI projects will be cancelled by the end of 2027 because of escalating costs, unclear business value or inadequate risk controls. Read Gartner’s project cancellation forecast

That suggests the harder part may not be building an agent.

The harder part is deciding where it should operate, how much authority it should receive and how the organisation will remain in control once it starts acting.

Indonesia is already moving in this direction from a policy perspective. The Ministry of Communication and Digital Affairs has stressed the need for accountability and human control as AI systems become capable of taking actions and is preparing national AI governance instruments.

Read Komdigi’s statement on Agentic AI governance

Technology Moves Fast. People Still Have to Answer for It.

There is a human capability problem underneath all of this.

Companies will need people who understand far more than how to use an AI application. They will need professionals who can test AI systems, manage access, investigate abnormal behaviour, audit decisions and determine where human approval must remain.

That is also why ITSEC established the ITSEC Cyber & AI Academy.

The objective is practical capability: preparing people who can work directly with cybersecurity and AI systems, understand how they behave and take responsibility for operating them safely inside real organisations.

Agentic AI will become easier to deploy.

Knowing when to let it act, how far to let it go and when a human must take over will be the harder skill.

Ready to build practical cybersecurity and AI capabilities?
Explore ITSEC Cyber & AI Academy

References

• Gartner, 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026
  Read the report
• Gartner, Hype Cycle for Agentic AI 2026
  Read the article
• Gartner, Over 40% of Agentic AI Projects Will Be Canceled by End of 2027
  Read the report
• Medcom.id, Techpresso coverage on Indonesia’s Agentic AI readiness
  Read Medcom.id
• Metro TV, Patrick Dannacher on access control for Agentic AI
  Watch the coverage
• Ministry of Communication and Digital Affairs, Indonesia
  Read the Komdigi statement