NTT DATA AIVista Closes the Last Mile of Enterprise AI
Critical steps for AI success in regulated industries.
1The Last-Mile Challenge in Enterprise AI
At VB Transform 2026, NTT DATA AIVista CEO Bratin Saha revealed why 70%+ of corporate AI projects fail during implementation. Poor integration, domain knowledge gaps, governance issues, and unclear ownership were cited as primary reasons. Saha emphasized that the system built around the model matters more than the model itself for delivering tangible business value.
2Insurance Industry Pain Points
Frontier models like GPT-5.5 show less than 60% accuracy in regulated workflows such as multinational insurance claims. Handwritten forms, complex checkboxes, and compliance requirements cause immediate failures. NTT DATA's solution implements customized guardrails that triple workflow efficiency while maintaining 98% compliance with 14 global regulations.
3Encoding Tribal Knowledge
Undocumented workflows containing competitive advantages become AI's most valuable input. By capturing institutional knowledge from insurance experts, NTT DATA developed agents delivering 40% more accurate outcomes. This knowledge transfer optimizes critical parameters like client risk classifications and regulatory interpretations specific to each enterprise.
4Cost-Effective Model Ensembles
The ROI of enterprise AI comes from surrounding systems, not base models. NTT DATA's ensemble approach reduces costs by 65% with only 2% accuracy trade-off. Strategic use of open-source models for low-risk processes saves clients $2.4M annually while reserving frontier models for high-stakes decisions.
5Change Management Hurdles
78% of employees resist radical workflow changes, making human factors - not technology - the primary adoption barrier. NTT DATA's phased transition strategy cuts adaptation time from 6 months to 3 while maintaining continuity in mission-critical operations. This approach decreases implementation failure rates by 55%.
6Automated Regulatory Compliance
In financial services, AI systems must comply with 14+ global regulations. NTT DATA's neuro-symbolic models perform automatic compliance checks, reducing legal violation risks by 92%. Custom guardrails trigger instant reprocessing when models err, maintaining audit trails required in heavily regulated industries.
7Scalable vs Bespoke Solutions
NTT DATA's platform scales guardrail generation while keeping knowledge modules client-specific. This hybrid model cuts deployment costs by 30% while tripling success rates. The company's 20-year experience as a top insurance administrator provides unmatched domain expertise in capturing institutional knowledge.
8The Future of Enterprise AI
By 2026, 80% of AI investments in regulated industries will focus on last-mile solutions. NTT DATA's workflow-centric approach delivers 4.7x higher ROI compared to model-centric deployments. Saha concludes: 'It's not about technology - domain expertise and change management discipline make the real difference in enterprise AI adoption.'
Original article: VentureBeat — NTT DATA AIVista ile Kurumsal AI'da Son Kilometre Çözümü