Day 1: AI Impact Summit 2026

Day 1: AI Impact Summit 2026 — Business and Execution Summary

Executive takeaways

• The summit’s central business message was that Indonesia needs to move from AI discussion and consumption to implementation: applying AI to productivity, industrial competitiveness, public services, MSMEs, and higher-value jobs.
• AI sovereignty was defined as capability and control, not isolation: Indonesia should be able to govern its data and AI deployment, operate technology, retain more economic value, and decide its own priorities—while partnering internationally.
• Speakers stressed that infrastructure alone will not create an AI economy. Adoption, skilled people, usable data, trustworthy governance, and solutions tied to real operational problems must develop together.
• The proposed ecosystem combines digital identity, compute access, AI-agent orchestration, local knowledge and language resources, research, and governance.

Strategy and operating priorities

Sovereign AI as a national capability

KORIKA co-founder and supervisory-board member Prof. Bambang Rianto framed AI as strategic infrastructure affecting economic competitiveness, resilience, and national sovereignty. He identified three priorities for the summit:

1. Deliver tangible public benefit.
2. Strengthen ethical, safe AI governance.
3. Accelerate adoption across industry.

Indonesia’s sovereignty agenda was described through four practical capabilities:

• Control and governance: Decide how and when AI is deployed, and how national data is used.
• Operational capability: Be able to run and maintain technology—not merely receive a technology transfer.
• Economic value capture: Retain more of the value generated by domestic digital services and platforms.
• Local relevance: Build AI that reflects Indonesian languages, culture, and national priorities.

Examples used to illustrate the risks of lacking operational control included a power plant that could not be operated locally after a technology transfer, and a telecom-related digital project that reportedly had to close after the operator could not work with the software source code. The speaker also cited an estimate that 95% of economic value from Indonesian digital apps and platforms is not captured domestically.

Government priorities for adoption and growth

Coordinating Minister for Economic Affairs Airlangga Hartarto argued that AI should raise productivity, strengthen MSMEs and industry, create higher-value jobs, and improve public services—not simply benefit large corporations. He identified four priorities:

• Digital infrastructure and energy: Expand data-center capacity and ensure reliable energy.
• Talent and innovation: Develop the workforce through education, scholarships, and industry training.
• Practical adoption: Apply AI in manufacturing, supply chains, logistics, taxation, health care, and public services.
• Governance and collaboration: Build trust, accountability, regulatory certainty, and international partnerships.

The minister emphasized that sovereignty should mean strengthening domestic talent, data, infrastructure, and technology while remaining open to investment and collaboration.

Frameworks, processes, and playbooks

• Research-to-impact pathway: Link research, shared infrastructure, industry partnerships, product testing, and deployment. BRIN described access to research facilities, funding schemes, mobility programs, and product-testing or certification support as parts of this pathway.
• Problem-first AI adoption: Identify a concrete operational problem before selecting a model or tool. Run a small pilot, test the hypothesis, learn quickly, and scale solutions that demonstrate value.
• Three AI workforce roles: KORIKA’s Muhammad Gifari described the need for a balance of:
  • AI takers who use AI in their work;
  • AI shapers who identify valuable problems and direct AI toward them;
  • AI makers who build models, tools, and systems.
• Reusable agent templates: Package AI-agent workflows, skills, and tools so they can be adapted and shared rather than rebuilt for each organization.
• Human-supervised trust: Test systems before deployment, monitor them in operation, provide escalation and complaint mechanisms, and increase an AI agent’s authority gradually as confidence grows.
• Modular governance and software architecture: Design systems so individual components can be changed as standards and regulations evolve, without rebuilding the whole system.
• Industrial AI adoption: Prioritize use cases with measurable gains in productivity, quality, resilience, or sustainability; pilot in factories or industrial parks, measure outcomes, then replicate successful solutions.
• Data-to-knowledge governance: Convert raw laws, regulations, and other trusted sources into structured knowledge that AI systems can use, helping reduce—but not eliminate—hallucinations.
• Regional risk response: A framework presented by AI Safety Asia highlighted six crisis-management elements: risk authority, information sharing, crisis authority, technical response, business continuity, and accountability.

Announced initiatives and products

• AI.id digital identity: PANDI launched the AI.id domain for businesses, startups, communities, and individuals. Speakers presented trusted, locally governed identity as a foundation for people, devices, services, and AI agents. PANDI said it manages 1.6 million .id domains and cited approximately 77 billion monthly traffic events across its infrastructure.
• KORIKA Cloud: An orchestration platform intended to connect AI models, providers, infrastructure, and users, rather than introduce another foundation model. Its aim is to make AI-agent development more accessible and support use cases in areas such as disaster response, health, and food security. Proposed agent templates could be reused, adapted, and shared by the community.
• KORIKA Token: A proposed rupiah-based AI-credit mechanism intended to broaden access to computing and connect users to Indonesian AI infrastructure. AI Capital said it planned to open a six-month community and ecosystem-shaping effort, with the initiative scheduled to go live on 15 November; details were expected to be finalized over the following two weeks.
• Compute deployment plans: AI Capital described plans to deploy around 2,000 inference engines from February, around 30,000 in April, and expand toward 300,000 over the following two years. It also said around 150,000 next-generation chips would be allocated to Indonesia in 2027–2028. The company said it had established an Indonesian chip-design subsidiary and aimed to recruit 100–150 local chip-design talent. These were company plans stated at the event, not reported completed deployments.
• KORIKA Project Nusa: KORIKA and AI Singapore launched an open-source collaboration focused on trusted AI foundations, governance, and local knowledge. The concept uses global models as generalists and regional resources—including Southeast Asian language models and local information—as specialists. The speakers highlighted human oversight, local relevance, and shared development as design principles.

Research, products, and use cases

BRIN’s recorded keynote described three research areas:

• Quantum machine learning: Research into possible applications of quantum algorithms, with an explicit need to benchmark them rigorously against conventional machine-learning methods. The speaker noted unresolved challenges in hardware, noise, scalability, and proving a genuine performance advantage.
• Post-quantum cryptography: Preparation for a future transition to quantum-resistant cryptography, including standards work, migration planning, workforce development, and coordination with the national cyber and cryptography agency. The keynote referenced a transition horizon around 2030.
• AI for cultural heritage: Using 3D scanning, computer vision, and biometric-style analysis to document and compare ancient sculptures. Proposed outputs include a digital catalog and tools to support artifact identification. The speakers stressed that AI-generated restorations and similarity scores should be treated as hypotheses and decision support—not as substitutes for archaeological judgment.

In the panel, BRIN also discussed:

• A research data repository, high-performance computing access, research funding, and researcher or student mobility.
• A geospatial platform combining satellite imagery for applications such as crop monitoring and disaster response. BRIN said it had about 10 applications and aimed to develop more than 30; it described potential access for public users, government, business, and research.
• Local-language data resources, including a corpus of more than two million language datasets and support for more than 700 local languages, as described during the discussion.
• Opportunities for industry and universities to use BRIN research infrastructure and data, including through partnerships and product testing.

Trust, safety, and organizational controls

Speakers repeatedly treated trust as an operational requirement for adoption:

• Test and monitor: Academic panelist Asril argued that evidence from testing, ongoing monitoring, and mechanisms for handling failure are all needed to support trust and certification.
• Control agent permissions: Panelists warned against giving autonomous agents unnecessary authority. One business example was an agent allowed to send emails; the recommendation was to restrict it to reading or drafting until the risks are understood and controls are in place.
• Scale permissions gradually: Treat agents like new employees: begin with limited authority, observe performance, and expand permissions only when justified.
• Keep humans accountable: Human oversight is especially important when AI can take actions through APIs, databases, email, or physical systems.
• Use adaptable standards: Singapore speakers described voluntary testing and guidance-based approaches, while noting that standards need to keep pace with changing technology. The business panelist advocated modular software architecture to accommodate regulatory changes.
• Prepare for cross-sector incidents: AI Safety Asia emphasized information sharing and continuity planning because AI failures may affect multiple sectors at once.
• Build local technical competence: AI Singapore’s William C. argued that countries need the technical ability to evaluate and align systems for local contexts—not just use them—and should invest in AI literacy as well as technical capacity.

Industrial AI and MSME adoption

UNIDO’s keynote focused on AI as an enabler of inclusive and sustainable industrial development. Suggested industrial applications included:

• Predictive maintenance to reduce breakdowns and downtime.
• Smarter production to improve productivity, quality, and competitiveness.
• Supply-chain resilience and industrial-park efficiency.
• Energy and water optimization, including the integration of renewable energy.
• Industrial symbiosis, where one company’s waste can serve as another’s input.
• Reliable environmental monitoring and decision support.

UNIDO stressed that weak input data produces weak outputs, and recommended investment in reliable data collection and skills. It also called for AI to be accessible to MSMEs, with support and hands-on assistance so smaller firms are not left behind.

The Ministry of MSMEs described a joint production-house / factory-sharing program, begun in 2022, with 16 co-production houses across 12 commodities. Its initial AI training covered commodities and businesses in South Sulawesi, Central Java, and West Java. The ministry’s two practical lessons were:

1. Assess digital literacy and business needs before designing the training or selecting tools.
2. Introduce AI where work already happens—in production, operations, or market access—rather than limiting use to chatbots that do not affect business processes.

Metrics and targets cited

Metric Figure or target stated Indonesia economic growth, Q2 2026 5.29% Indonesia growth, first half of 2026 5.45% Government growth target, 2027 5.8–6.5% Government growth target, 2029 8% Indonesia digital economy, projected for 2026 US$100–140 billion Indonesia AI economy, potential by 2030 US$360–400 billion Global AI investment, projected for 2026 More than US$1 trillion Digital talent needed by 2030 About 9 million Semiconductor design engineers: Danantara–ARM training aim 15,000 within three years Internet users cited by the minister 229 million Digital MSMEs cited by the minister 22 million Data-center electricity use cited 485 TWh in 2026, rising to 940–950 TWh Electricity capacity and solar programs cited 70 GW in the 2025–2034 power plan; 100 GW solar program Large enterprises using AI, ministry estimate 62.5% Small and medium businesses using AI, ministry estimate 14.5% AI pilots failing / succeeding, cited in the industry panel 95% / 5% Executives making AI a strategic priority / realizing value, cited in the industry panel 80% / 25%

These figures were presented by speakers and were not independently verified in the subtitles. The data-center energy figure was stated as transcribed; its units and wording may be inaccurate.

Actionable recommendations

• Begin with a measurable business or public-service problem, not a technology purchase.
• Use small, controlled pilots; define success criteria and operational ownership before scaling.
• Build data foundations and domain-specific knowledge, particularly for local languages and sectors.
• Make research facilities, datasets, and testing support accessible to universities, startups, and industry.
• Train people not only to use AI but also to define problems, evaluate outcomes, build systems, and supervise agents.
• Restrict AI permissions by default and expand them in stages, with monitoring and clear accountability.
• Prioritize adoption across MSMEs and public services, not only investment in large-scale compute.
• Combine domestic capability-building with international partnerships; avoid both technological dependence and impractical attempts to build everything alone.

Presenters and sources

The event was hosted by KORIKA as part of Bali Techweek 2026. Speakers and moderators featured in the subtitles included MC Getri; Prof. Bambang Rianto; Dr. Edwin Hidayat Abdullah; Prof. Arif Satria (recorded keynote); Coordinating Minister Airlangga Hartarto (virtual keynote); PANDI chair Isnawan Aslam; KORIKA’s Prof. Hamam Riza, Dr. Muhammad Gifari, and Oscar Riandi; AI Capital Indonesia’s Prof. Niklas Adler; AI Singapore’s William C.; AI Safety Institute Indonesia’s Nur Anis Hadiati; Asril (academic panelist); AI Safety Asia’s Devi; UNIDO Chief Technical Advisor Salil Dutt; and panelists Lydia Gandini Triastuti, Chan Wichi, Kiril Ignatkov, and Bagas Wicaksono. The final industrial panel was still in progress when the subtitles ended.