Lessons for India from Estonia, world's 1st digital society & lessons for India

Technology-focused summary (from the subtitles)

Core requirement: Data governance + shared governance architecture

• Estonia’s key obstacle before fully digitizing services was data governance—digitization required multiple organizations to coordinate.
• Major shift: building data ownership, plus rules for data quality, accessibility, and findability.
• Evidence of scale-up: Estonia’s open data portal grew from <50 datasets (around 2018) to >8,000 datasets later, attributed to governance and architecture work.

Trigger for faster digitization: COVID-driven changes

• Estonia accelerated digitization when activities typically done in person began changing (e.g., interactions that involved notaries for real estate transactions).
• It introduced/expanded personalized proactive services, integrating multiple government organizations to handle citizen life events (e.g., marriage → divorce workflows).

Inclusive digital society: accessibility + real-time content adaptation

A successful digital society should support:

• Citizens without advanced ICT skills using analog/alternative channels
• People with hearing or sight impairments

During COVID:

• Estonia built a real-time subtitling solution for national broadcasting.
• It extended the approach to other sectors (e.g., theaters/private sector) to ensure equal access.

After Russia’s attack on Ukraine:

• Estonia implemented Estonian–Ukrainian translation integrated into service portals for refugees to help them access critical services.

Digital governance success criteria (citizen-centered)

• Technical complexity should be “hidden” from individuals.
• Citizens care about:
  • Their obligations and opportunities
  • What data the government processes
  • How trust is ensured
• Emphasis on consent-based data sharing.

Transferable principles for India (at different scale)

• Estonia-to-India replication is not about copying systems, but applying principles, such as:
  • Data accessibility requirements
  • Data quality assurance
  • Human/citizen control
  • Digital-first approach as a universal method
• Scale difference (1.3 million vs 1.3 billion) affects implementation, but the governance principles remain applicable.

AI’s impact: from “assistance” to “agentic” and predictive governance

Estonia metrics mentioned:

• >50% of Estonians use AI
• >80% of young people use AI
• Government adoption around 37% (sector-dependent)

Key idea:

• Citizens increasingly use AI agents to interact with government; government interactions are becoming agentic.

Examples:

• Unemployment agency
  • Shift from “gut feeling” recommendations to prediction, including:
    • best job match
    • training needs
    • likelihood of finding employment
  • Next step: prevention and personalization—anticipate risk and intervene early.
• Health sector
  • Shift described as moving from curing to ensuring people stay healthy, implying predictive/personalized services.

Governance/architecture requirements with AI:

• Sovereignty
• Human control over disputes
• Audibility/explainability (visibility into what agents do)

Collaboration model: Estonia’s public-private partnership

• Estonia used public-private collaboration:
  • Government ensures trusted data sharing, assurance, compliance, and expectations.
  • Private sector builds services (e.g., healthcare, transportation), improving cost-effectiveness and scalability.

Key product feature: Consent service (developed starting 2020)

• Gives individuals the right to decide who can access which government data and for what reason.
• Example: applying for a bank loan
  • Consent enables sharing employment/salary information so banks can respond faster and avoid repetitive forms.
• Also applied in healthcare
  • Consent enables additional services by private actors using government data.

Lessons/general guidance for India and others

• Avoid over-complexity: “keep it simple.”
• Start small, test, and iterate:
  • If a solution doesn’t work, scrap/adjust
  • Focus on minimum scope that delivers user value
• Be cautious about “AI for everything”:
  • Use AI only after identifying the real problem/challenge through service and user design
• AI sovereignty + cultural representation:
  • Ensure systems can represent local language and culture, not just global vendor interests
  • Use concrete, measurable objectives (mentioned: 160+ AI projects with focused impact)

---

Main speakers / sources

• Akanksha Mishra (The Print, interviewer)
• Dr. Ott Velsberg (former Chief Data Officer of the Estonian government; Tony Blair Institute for Social Change)