
Can AI cut the time Indonesians wait for public services? The Deputy Minister of Communication and Digital Affairs, Nezar Patria, has asked every government unit to find out. Speaking online at an AI training session on 30 October 2026, he asked each unit to pick at least one service where AI could shorten waiting times from weeks to days before the end of the fourth quarter of 2026. On 5 November, at Universitas Gadjah Mada, he added that regional governments should share AI components rather than each build their own. For companies that apply for licences, submit documents or supply technology to government, the push is worth understanding, and its open questions matter as much as its ambition.
What the deputy minister asked for
His remarks were reported by InvestorTrust and Niaga.Asia on 31 October, RRI on 1 November, and JawaPos and Berita Daerah on 2 November. Niaga.Asia credits the ministry as its source. The session was titled AI Value Chain: Indonesia's Strategy in the AI Supply Chain as a Middle Power, and he took part online from Central Jakarta.
- The target: each unit should name at least one service where AI could cut the wait from weeks to days before the quarter ends. Berita Daerah words it as several weeks to several days.
- The kind of work: document processing, licensing, information services for citizens and regulatory compliance. None of the reports we read names a specific service.
- Beyond pilots: he said AI should not stop at demonstrations or pilot projects that leave the quality of service unchanged.
- The competitive point: agencies that adopt AI may pull ahead of those still on older systems, and private firms using AI could handle documents and licensing faster than the bodies that hold the authority. He described not adopting AI as not a reasonable choice.
- The method: understand what the public needs, find the obstacles in service processes, then decide which tasks technology can speed up. Berita Daerah adds that units should set a clear problem, a measurable target and a short time frame.
Shared components for 508 regions
On 5 November, at the Gadjah Mada Digital Transformation Governance Index 2026 event at UGM in Yogyakarta, Nezar argued that Indonesia's 508 regencies and cities need not each build a similar system, or in his words, their own 508 chatbots, Warta Ekonomi reported on 6 November. He cited Estonia, which he said had consolidated many chatbots, and named language services, complaint modules and data standards as components that could be shared, so that regional budgets go on solving problems rather than repeating work.
He also proposed an AI Public Service Lab, bringing together students, lecturers, civil servants, businesses and communities. The process he described is to identify a problem, design a solution, run limited trials or sandboxing, measure the impact, then expand in a measured way. He said success should not be judged by adoption alone, and that the benefits must reach people at kelurahan and kecamatan level. The index, developed by UGM, now covers 508 regencies and cities, up from 429 regencies two years ago, and has 7 pillars and 50 indicators.
How this fits earlier statements
The challenge follows two months of related remarks. At the AIPS Demo Day in Jakarta on 15 September, reported by the Satu Data Indonesia portal, Nezar said the main measure of AI in government is whether citizens feel the government has become more capable, responsive, inclusive and trustworthy. AIPS, the AI Incubation for Public Sector programme, opened in May 2026 and received more than 170 registrations for five government challenges. Five teams then worked with Bappenas, the Ministry of Finance, BPJS Kesehatan, the Ministry of Primary and Secondary Education and the Ministry of Investment and Downstreaming until September. At the same event, Dini Maghfirra, a director at Bappenas who also leads the Satu Data Indonesia secretariat, said technology is no substitute for governance, and that differing data standards between agencies hinder interoperability.
In a statement dated 21 September for the IABC Indonesia conference, reported by InfoPublik, Nezar warned that errors in government AI systems can affect citizens directly and weaken trust in institutions. He called for safeguards, testing, sandboxing, compliance and strict evaluation before wide use. Separately, Katadata reported on 7 October that a national AI roadmap for 2026 to 2029 is being drafted with more than 30 ministries and agencies, with bureaucratic reform among five priorities. We found no report that it has been issued.
What is not yet clear
- Measurement: the reports do not say how a cut from weeks to days will be measured, or whether units must report the service they choose.
- Status: the target was set out as a challenge at a training session. We found no written circular or regulation behind it.
- Shared components: no report names a platform, a budget, a timetable or who would build and run them.
- Safeguards: none of the reports on the quarterly target explains how the testing and evaluation he has called for before would fit into a single quarter, and none mentions personal data protection or human review.
What organisations can prepare
This section is our analysis, not legal or procurement advice.
- Businesses that rely on licences and permits: watch which services agencies choose. Faster, AI-assisted channels may change document formats and submission steps, so keep company documents complete, consistent and machine-readable.
- Technology suppliers: the stated test is a measurable change in a short time, and the stated preference is shared components over separate systems for each region. Proposals that state the current waiting time and how improvement will be measured fit that test.
- Teams building AI for documents or licensing: build in an evaluation set, a record of what the system did and a route to a human officer for unclear cases, in line with the safeguards the deputy minister has named.
- Data owners: the point made at Bappenas applies outside government too. Data exchanged with agencies is easier to use when it follows common standards.
- Private enterprises with stalled AI pilots: the same discipline applies. Pick one process, record its baseline, set a deadline and measure the result.
What to watch next
Watch whether Komdigi publishes which services units chose and what changed by the end of December. Watch whether the shared components and the AI Public Service Lab are given an owner, a budget and a timetable. And watch for the national AI roadmap for 2026 to 2029, and for whether the safeguards the deputy minister has described, testing, sandboxing and evaluation, are written into guidance for agencies.
Sources
- Nezar Patria Tantang Instansi Pemerintah Gunakan AI untuk Percepat Pelayanan Publik · InvestorTrust(opens in a new tab)
- Nezar Patria Tantang Instansi Pemerintah Pangkas Layanan dari Minggu Jadi Hari dengan AI · Niaga.Asia(opens in a new tab)
- Wamenkomdigi: Kecerdasan Buatan Harus Pangkas Waktu Pelayanan Publik · RRI(opens in a new tab)
- Komdigi Tantang Instansi Pemerintah Pangkas Waktu Pelayanan Pakai AI · JawaPos(opens in a new tab)
- Wamenkomdigi Dorong AI Pangkas Waktu Pelayanan Publik · Berita Daerah(opens in a new tab)
- Wamenkomdigi Nezar Sebut Daerah Tak Perlu Bangun Sistem AI Sendiri · Warta Ekonomi(opens in a new tab)
- AI di Pemerintahan Harus Berorientasi pada Dampak dan Kepercayaan Publik · Satu Data Indonesia (data.go.id)(opens in a new tab)
- Kepercayaan Jadi Fondasi Pengembangan AI di Indonesia · InfoPublik(opens in a new tab)
- Komdigi Minta Proyek AI Cetak Talenta dan Pemasok Teknologi Lokal · Katadata(opens in a new tab)
This briefing was prepared by the Indotek editorial desk with AI assistance, from the public sources listed above. It is general information, not legal or professional advice.
AI & Agentic AI
How Indotek can help
Cutting a document or licensing process from weeks to days, without losing control of the decisions in it, is a problem of document handling, workflow and evaluation. Designing and building that kind of system is the work Indotek does in AI.
- Document intelligence: we build extraction, classification and validation for forms, letters and supporting documents, with confidence scoring and human review for the cases that need it.
- Workflow automation: we combine deterministic rules with AI judgement so that exceptions in a licensing or service process are handled rather than dropped.
- Evaluation before scale: every AI feature we ship comes with an evaluation set, so that a claimed cut in waiting time is measured rather than assumed.
- Humans in command: our agentic systems act inside explicit permissions, log what they do and hand off to a person when confidence is low or the stakes are high.



