High-level discussion on APAC Centre for AI

Detailed Summary

• Moderator (Kamesh Shekar) opened the session, thanked the audience for turning up despite a “chaotic” schedule, and explained that the hour would be split into a panel discussion followed by an open audience dialogue.
• He outlined the origin of the APAC Centre idea: Core AI, a large Indian multi‑stakeholder AI coalition, had partnered previously with Plug and Play and GiveCity and saw the need for a regional hub that could address Asia‑Pacific‑specific challenges.

2. Panelist Introductions (quick roll‑call)

Panelist	Core Affiliation	Primary Perspective
Shweta Tiwari (Plug and Play)	Global accelerator with APAC focus	Startup‑centric opportunities & ecosystem roadblocks
Dr. Sivaramakrishnan Guruvayur (Aquarians AI)	AI research‑startup	Applied AI use‑cases, sectoral traction
Jeremy Fritzhand (Ahmedabad University)	Venture‑studio/technology incubator	Soft‑landing programmes, academic‑industry bridges
Mohit Jain (Microsoft Research)	Research in regulated domains (healthcare)	Technical interoperability & data‑trust
Geeta Kanwar (Caria AI)	Policy & government engagement	Governance, data‑quality, evaluation frameworks
Varun Sakhoja (Mastercard India)	Payments & corporate‑public partnership	Institutional pilots, awareness & trust
Mayuresh Ramachandra (Salmander Advisors)	Deep‑tech fund	Funding gaps, investment landscape
Praveen Das (Wikimedia Foundation)	Open‑knowledge & multilingual data	Open‑source datasets, digital public goods

Note: Several additional experts (Aditi Sita, Sonia Bhaskar, Tripti Bansal) were introduced as open‑discussion facilitators and would join later.

3. Key Themes from the Panel

3.1 Opportunities & Roadblocks in the AI Ecosystem (Shweta Tiwari)

• Fintech analogy: The early‑stage panic in 2014‑15 (banks “vanishing”) mirrors current AI anxieties; the outcome is greater efficiency, not massive job loss.
• Start‑up focus: Modern AI start‑ups are not only building novel algorithms but re‑engineering existing organizational workflows with AI‑enhanced processes.
• Scale‑up mechanisms: Plug and Play runs 600+ programmes globally, positioning itself to aggregate talent and connect corporates with AI solutions.
• Roadblocks:
  • Talent concentration: need for a “center” to pool AI talent across APAC.
  • Capital access: early‑stage deep‑tech ventures struggle to secure seed funding due to a “wait‑and‑watch” mentality among VCs.

3.2 Startup Viewpoint & Sectoral Demand (Dr. Sivaramakrishnan Guruvayur)

• Applied‑AI over pure tech talk: Customers now demand business impact—“What problem are you solving?”—instead of generic generative‑AI hype.
• Hot sectors:
  • BFSI (Banking, Financial Services, Insurance)
  • Healthcare (especially regulated data pipelines)
  • Public services / government (complex workflow automation)
• Challenges:
  • Capital scarcity – VCs are cautious after recent market turbulence.
  • Match‑making – Start‑ups need structured pipelines linking them with use‑case owners (corporates, ministries).
• Strategic advice: Deep‑tech combined with domain expertise is essential for scaling.

3.3 Cross‑Border Collaboration & “Soft‑Landing” Programs (Jeremy Fritzhand)

• Soft‑landing definition: Programs that help a start‑up enter a new market by providing regulatory guidance, funding navigation, and partner matchmaking.
• Reciprocity principle: India has historically received more than it gives; future programmes must be mutually beneficial (e.g., shared fund‑index, joint scouting).
• Policy levers:
  • Easier EXIM policies (Export‑Import) – recent EU‑India and UK‑India agreements are promising.
  • Tariff reduction & trade‑agreement improvements – vital for AI‑focused SMEs.
• Academic role (Jeremy’s follow‑up):
  • Muntun platform (led by Dr. Sapnapoti) indexes public‑private AI pilot opportunities.
  • Suggests adding commercialization metrics to university rankings (start‑ups spun out, licenses filed).
  • Push for entrepreneurship education from school age, aligning AI talent pipelines with national priorities (healthcare, agriculture, Indian Knowledge Systems).

3.4 Corporate‑Public Partnerships (Varun Sakhoja, Mastercard)

• Mastercard “Start Path” analogy: A global platform that selects 200 partners across 210+ countries, runs structured pilots, and delivers AI solutions to regulators, governments, and merchants.
• Public‑private synergy:
  • Start‑ups need technology, models, best‑practice access.
  • Corporates need implementation partners (start‑ups) for rapid roll‑out.
  • Governments require direction & standards – a centre could orchestrate these three.
• Awareness gap: Even in “Tier‑2/3” cities, basic cybersecurity hygiene (e.g., not sharing OTPs) is lacking; an APAC centre could run regional awareness programmes.

3.5 Technical Interoperability & Data Trust (Mohit Jain, Microsoft Research)

• Regulated domains concern: Healthcare, finance, and legal sectors hold sensitive data; cross‑border sharing is hindered by trust deficits.
• Policy need: Clear, region‑wide data‑use policies that reconcile privacy with innovation.

3.6 Governance, Data Quality & Evaluation Benchmarks (Geeta Kanwar, Caria AI)

• Data representativeness: Existing language datasets are Eurocentric; APAC’s 70+ Indian languages are under‑represented, leading to bias at the embedding stage.
• Evaluation frameworks: Current safety matrices ignore low‑resource settings and culturally specific bias.
• Recommendation: The centre should construct APAC‑centric benchmarks that test:
  • Model performance on low‑resource languages.
  • Cultural relevance and region‑specific bias mitigation.

3.7 Decentralised‑Centralised Governance Model (Unattributed audience contribution)

• Hybrid approach: 60 % of the governance framework could be common across APAC, with local variations to accommodate jurisdictional nuance.
• Consensus on “high‑risk AI”: Need a shared definition before regulation.

3.8 Open‑Source Data, Multilingual Internet (Praveen Das, Wikimedia)

• Wikimedia as a digital public good: Provides 300+ language editions; its open‑governance model offers transparent provenance, key for trust.
• AI‑training datasets: Open, multilingual corpora (e.g., Wikidata) should be easily accessible for cross‑border collaboration.
• Policy implication: Encourage open‑source data‑set repositories (e.g., India’s “AI‑Kosh”) to underpin shared model development.

3.9 Funding Landscape & Investment Gaps (Mayuresh Ramachandra, Salmander Advisors)

• Three pillars of growth: Labor, capital, productivity. India must boost productivity via AI‑enabled gig‑work and enterprise efficiency.
• Capital realities:
  • Cannot replicate the US deep‑pocket model nor the China state‑funded model.
  • Must develop a mixed‑funding ecosystem (government, corporate, limited‑partner funds).
• Sovereign LLMs: Emphasis on language‑specific large models to serve Indian and broader SE‑Asian markets.
• Trust deficit: Deteriorating global trade structures have eroded trust; the centre should act as a trust‑broker.
• Investment efficiency: Existing Indian deep‑tech funds (e.g., C‑Fund) highlight a “transactional” culture—lack of collaboration leads to waste. The centre should facilitate knowledge‑sharing among IITs, IIMs, and other research bodies.

3.10 Trust & Interoperability (Mahirish – audience member)

• Payments analogy: Trust built over decades in payments (e.g., crypto‑free, cross‑border settlement) enables interoperable transactions.
• AI‑trust framework: Proposes a federated‑learning model for secure, privacy‑preserving cross‑border AI collaboration.

4. Open Audience Dialogue (Expert‑led Q&A)

Expert	Core Point(s)	Suggested Action for the APAC Centre
Sonia Bhaskar (Public‑information integrity)	Responsible AI must be inclusive; current datasets lack gender, sectoral representation (e.g., women in healthcare).	Embed data‑inclusion mandates; run digital‑AI literacy programmes.
Tripti Bansal (Policy Index)	APAC’s cultural & linguistic diversity demands non‑uniform frameworks; success‑playbooks should be regionalised, not “Western”.	Publish contextual case studies, guide replication across heterogeneous markets.
Aditi Sita (Dutient)	Existing AI regulations are fragmented; start‑ups face a policy‑maze when scaling across borders (e.g., data‑localisation, explainability).	Act as a policy harmoniser and do‑tank: issue standardised risk‑assessment templates, provide compliance “hand‑holds”.
Mahirish (Audience) – brief interjection	Trust is the foundation; similar to payment systems, AI needs a cross‑border trust layer.	Develop a federated‑learning based trust framework and a sandbox for testing.
Ramashiv Kumar (SRK Game Changers)	Need a culture that tolerates failure; deep‑tech incubation should encourage learning from setbacks.	Centre should host failure‑postmortems and mentor networks.
Vidit (Adobe)	Accessibility for under‑served populations (farmers, villages).	Prioritise low‑bandwidth AI tools, community‑centric deployments.
Amit Mithil (Aviation)	Jurisdictional risk when AI intersects with safety‑critical domains (aviation, civil aviation).	Create a legal‑risk repository and cross‑border jurisdiction guidelines.

All experts emphasized a do‑tank approach: the centre must move beyond policy‑paper to pilots, sandboxes, and concrete tooling.

5. Closing Remarks

• Moderator (Kamesh) thanked participants, acknowledged the time constraint, and promised follow‑up outreach to capture deeper inputs from experts who could not fully speak.
• He reiterated that the discussion was only a starting point for a regional collaborative architecture that will evolve through continuous stakeholder engagement.

Key Takeaways

• APAC‑wide AI hub is needed to pool talent, align standards, and accelerate sector‑specific AI adoption across a culturally and linguistically diverse region.
• Start‑ups thrive on applied AI: they must focus on measurable business impact rather than generic generative‑AI hype.
• Cross‑border “soft‑landing” programmes should be reciprocal, offering both funding insight and market‑entry assistance to all participating countries.
• Data quality and representativeness are critical; the centre must drive APAC‑centric language datasets and culturally aware evaluation benchmarks.
• Policy harmonisation is essential: a unified, yet locally adaptable, AI regulatory framework will reduce friction for start‑ups expanding across borders.
• Public‑private‑government partnerships (e.g., Mastercard’s “Start Path” model) illustrate how structured pilots can deliver tangible AI outcomes for regulators and citizens.
• Trust is the linchpin: an AI‑specific trust framework—potentially built on federated learning and sandbox environments—will underpin cross‑border collaboration.
• Funding gaps demand a mixed‑source capital model and better match‑making between deep‑tech founders and investors; the centre can act as a catalyst.
• Inclusivity and accessibility must be baked into every initiative, ensuring that AI benefits reach underserved populations (rural farmers, low‑resource language speakers, etc.).
• Governance should blend centralisation and decentralisation: a core APAC standard with local variations can address high‑risk use‑cases while respecting sovereignty.

---

Prepared from the verbatim transcript of the “High‑level discussion on APAC Centre for AI” panel at the India AI Impact Summit 2026.

See Also:

• india-japan-ai-for-the-world
• ai-innovators-exchange-accelerating-innovation-through-startup-and-industry-synergy
• building-resilient-sustainable-ai-infrastructure-for-people-planet-and-progress
• high-level-panel-discussion-on-ai-bridges-indiaisrael-innovation-dialogue
• the-sustainable-digital-infrastructure-accord-driving-sustainability-of-ai-infrastructure-in-the-asia-pacific-region