Manish Rai of SnapLogic highlights how companies who are successful with GenAI are those that have identified opportunities, managed risk, and built for scale
GenAI has dominated global headlines, boardroom discussions, and investment strategies since ChatGPT’s debut two years ago. The technology’s potential is undeniable: automating complex tasks, creating new customer experiences, even unlocking entirely new business models, with organisations who have AI-led processes outperforming their peers. Yet, despite the noise and billions of dollars investment, businesses remain stuck.
A 2022 Gartner survey revealed that just 54% of AI projects make it past the pilot stage, while a 2024 year-end report from Deloitte reports that nearly 70% of GenAI experiments never reach production. While the possibilities are vast, practical success remains out of reach. We’re seeing a growing disconnect between proof-of-concept and actual production-level execution. The question is no longer whether GenAI works, it’s whether businesses are ready to make it work.
Why promising GenAI projects stall
The final hurdle is not technical capability but organisational readiness. Too many projects launch with abstract goals, disconnected from operational priorities or measurable ROI. Successful GenAI use-cases are still new, so most enterprises struggle to identify and prioritise them.
The success of GenAI also heavily depends on large, high-quality datasets. However, most enterprises lack the infrastructure, integration or data-hygiene to provide this input at scale. Without usable data, even the most powerful GenAI models will remain theoretical. Most organisations and their datasets were not designed with GenAI in mind, leading to data siloes and difficulties in accessing datasets needed for effective GenAI performance.
Governance concerns are becoming increasingly important for businesses too. As enthusiasm rises, so does leadership anxiety around security, bias, compliance and unintended consequences. Too often, these issues are deferred or poorly understood, and by the time a project nears production, uncertainty around risk and a lack of framework risks putting a stop to otherwise successful projects.
Add to this a lack of internal playbooks – very few organisations have standardised processes for identifying and prioritising use cases, determining when to use GenAI, enforcing data security and privacy policies, and evaluating AI performance. This can leave businesses with an AI project that is too fragile to scale altogether.
“Successful companies understand that deployment isn’t a technical milestone, it’s an organisational commitment”
Keys to fulfilment
So, let’s look at what successful enterprises do differently in order to ensure that their GenAI visions don’t wither on the vine:
Modernise integration and data strategy: Legacy systems cannot support the speed and complexity of GenAI and its demands. Organisations seeing success have invested in modern integration platforms capable of handling structure and unstructured data, API integration and management, as well as third-party applications – all within a flexible, composable architecture. These platforms not only accelerate experimentation but ensure scalability when the time comes to deploy.
Embed governance from the beginning: Instead of treating governance as a post-launch concern, successful teams integrate security, privacy and compliance into every stage of GenAI development. Clear oversight frameworks, approval checkpoints, and risk assessments are embedded into workflows, which avoids the last minute-panic that so often derails deployment.
Create GenAI centres of excellence: To avoid fragmentation, many organisations are creating cross-functional Centres of Excellence that combine technical, legal and business expertise. These teams are responsible for developing best practices, aligning projects with business strategy and accelerating enterprise-wide adoption. Their presence signals a long-term commitment not just to AI, but the operating model that is required to make it sustainable.
Bridge the skills gap: The rapid evolution of AI requires continuous learning. Rather than wait for the perfect hire, leading companies are investing in upskilling existing talent, whilst also bringing in external partners who can share frameworks, accelerate timelines, and reduce early-stage risks. This hybrid approach ensures institutional knowledge is preserved, whilst benefitting from cutting-edge expertise that doesn’t overwhelm internal resources.
From obstacles to transformation
Barriers preventing successful GenAI deployment are nothing new. Poor data quality, unclear governance, and talent shortages are the same foundational issues that have challenged digital transformation for years. What GenAI is doing now is enforcing a reckoning, exposing the gaps in integration, AI strategy and the very design of businesses. Through exposure comes opportunity, not just to deploy GenAI, but to modernise how businesses are run.
Successful companies are not treating GenAI as just another IT initiative. They are treating it as a catalyst for broader operational and industry change. They understand that deployment isn’t a technical milestone, it’s an organisational commitment. These companies are building the capabilities, culture, and clarity needed to sustain it.
To avoid being on the wrong side of the 54%, business leaders must rethink how they identify opportunities, manage risk and build for scale. The infrastructure, talent, and use cases of GenAI are increasingly becoming understood. Now, companies must come up to par.
About the author
Manish Rai is VP of Product Marketing at SnapLogic.
Further reading
This article was first published in issue 2 of Business 4.0.