MIT’s GenAI Divide report has made waves with one stark finding: 95% of enterprise GenAI pilots show no measurable P&L impact within six months.
At first glance, the verdict looks grim - but it’s not the technology that’s failing. It’s the way pilots are being run. Too many get stuck in “innovation theatre”: slow data approvals, misaligned vendors, and proof-of-concepts that drift for months without defined outcomes.
At NayaOne, we see this pattern firsthand. Banks and insurers launch dozens of AI experiments every year. Without structured validation, most stall. With the right infrastructure, measurable impact can emerge in as little as 90 days.
Beyond the “95% Fail” Headline
MIT’s definition of “failure” is narrow: no quantifiable revenue or cost benefit within six months. For financial institutions, where innovation cycles span 12 - 24-months, that’s often unrealistic. The real lesson is that most GenAI projects don’t transition from pilot theatre to production.
This aligns with what we see in client PoCs:
- Data access delays: Institutions often wait 4 – 6 months for data approvals before testing can begin. With synthetic data libraries, NayaOne reduces this to days.
- Vendor misalignment: In over half of PoCs, the first vendor evaluated doesn’t meet compliance or integration requirements. Structured shortlisting prevents wasted onboarding.
- Drifting pilots: Without defined exit criteria, PoCs run 9 – 12 months without outcomes. On NayaOne, clients decide to scale or stop in under 90 days.
From the Field: Claims Automation
A global insurer tested GenAI models for claims triage and fraud detection in NayaOne’s sandbox. Within 12 weeks, they identified bias in vendor outputs and validated integration with legacy claims systems. Instead of stalling in proof-of-concept theatre, they moved forward with two vendors - and avoided onboarding three that would have failed regulatory checks.
Why Integration Breaks Down
MIT’s research highlights integration - not AI quality - as the bottleneck. Banks face:
- Governance friction – Risk and compliance approvals delay testing.
- Workflow misfit – Models don’t align with legacy processes.
- Unclear ROI – Success metrics are rarely defined up front.
Our clients’ experience confirms this. For example, one Tier 1 bank tested four fraud detection vendors in parallel. Using NayaOne’s sandbox and synthetic data, they surfaced two with measurable accuracy gains within 90 days - cutting six months of onboarding effort and saving ~$500k in internal validation costs.
From the Field: KYC Automation
A European bank validated AI-driven ID verification vendors using NayaOne’s synthetic customer data. By surfacing audit and integration concerns early, they reduced data-approval cycles by 70% and cleared governance checks before vendor onboarding began.
Implications for Financial Services
- High failure risk without structure: Projects drift in pilot theatre and never impact P&L.
- Trust and compliance are at stake: Bias in lending or fraud detection models creates reputational and regulatory exposure.
- Pressure to show ROI is growing: $44B invested in AI startups in H1 2025 means executives are expected to deliver results quickly.
Recommendations for Leaders
Redefine success horizons: Track productivity, risk reduction, and cost efficiency in the first 6 – 12 months, not just revenue.
Kill pilot theatre: Every PoC should have defined ROI metrics and a path to scale.
Adopt validation infrastructure: Use sandboxes, synthetic data, and compliance workflows to test securely before onboarding.
Balance portfolios: Avoid overconcentration in AI; diversify across fraud, identity, payments, and operational efficiency.
Maintain human oversight: Position AI as augmentation, not replacement, in customer-facing and regulated workflows.
Four winning patterns we see from the field:
- Run more proof-of-concepts. Like VC math, assume many won’t deliver but the few that do can transform the business. Diversify experiments so you don’t bet everything on a single initiative.
- Intense bursts by small teams. Hackathons and tiger teams move faster and validate use cases before scaling into the wider organisation.
- Go top-down and bottom-up. Leadership sets focus, governance, and trust, while front-line teams drive experimentation and surface practical use cases. Both are essential.
- Continuously learn and adapt. Treat AI adoption as a series of sprints, with each iteration refining models, processes, and ROI.
The MIT study isn’t a verdict on AI’s potential. It’s evidence that without structured validation, enterprises will continue to see little measurable ROI.
NayaOne’s Vendor Delivery Infrastructure (VDI) changes the odds by providing:
- ROI tracking frameworks that prevent drift and ensure outcomes. to test integrations safely.
- Synthetic data libraries to accelerate compliance.
- Vetted vendor marketplace to avoid misaligned suppliers.
- ROI tracking frameworks that prevent drift and ensure outcomes.
