By Ed Granger
Enterprise generative AI is moving beyond the hype phase. Businesses across a range of industry verticals are experimenting with the technology in pilot programmes and tiger teams.
In truth, this represents somewhat of a crossroads rather than a jump-off point into AI nirvana with the much-heralded benefits that brings. Indeed, a recent report by Alteryx found that despite 74% of generative AI pilot projects proving successful, over one-quarter of businesses are still struggling to integrate the technology.
So, what’s behind the slow pivot from AI pilot programmes to enterprise-wide rollouts? It’s fair to say that it’s not user appetite. Last year, a KPMG survey found 61% of British workers want specific training on how to use the technology.
However, channelling interest into practical workflows is difficult. And it appears one of the key challenges arises from integrating generative AI in existing organisational structures, strategies and workflows.
There are plenty of ways for AI rollouts to become disjointed, leading to suboptimal outcomes and missed opportunities for true transformation. A Retail AI Council and Salesforce study published in March was telling. It found nearly half of 1,300 global retailers are struggling to make their data accessible for generative AI models. This is one clear example of a foundational element for generative AI success that many are finding difficult.
To avoid such issues, the right set of priorities is needed throughout the process of rolling out generative AI – that includes the very initial stages of planning.
Putting the best foot forward means recognising scale
Before diving into specific use cases, organisations must go on the important journey of assessing their readiness for AI adoption. Who will be the internal stakeholders using new generative AI applications and what’s their level of understanding of the technology? Does the organisation have an internal culture that’s welcoming of experimentation and innovation and, if so, likely to welcome new AI technology? Are there knowledge gaps on the ethical and governance considerations that need to be factored into AI rollout? And are the data, technology platforms and internal resources available to actually deliver on AI’s promise?
If it wasn’t obvious already, this assessment has to be wide-reaching. This isn’t just a case of implementing a technology. It’s developing and fostering an organisational culture that embraces AI-driven changes. This requires an effective strategy for managing and adopting AI across an enterprise.
The good news is that more C-suite leaders today can see the need for that strategy and are pulling the right team together to manage and execute. As enterprises have shifted further and further to digitisation, we see more COOs and Chief Strategy Officers hailing from technologist backgrounds. Many are finding it prudent to empower enterprise architects (EAs) – those in an organisation typically responsible for designing and planning enterprise analysis – to play a major role in delivering AI transformation.
EAs are the AI transformation shepherds
EAs are well-placed to support and manage the essential steps for successful generative AI. Building an effective data pipeline, for example, is an important step. However, it was identified by Gartner in a recent ‘board brief’ on generative AI as a resource-intensive task that requires a strong focus on various activities, including data integration and optimised data structures, data quality and security and, finally, governance.
In other words, it needs a holistic approach.
Today’s EAs are already familiar with gathering information from across the organisation and generating insights to inform the evolution of business processes and technology landscapes. So when it comes to AI, EA’s overarching role provides unique value in mapping the end-to-end business landscape to identify AI innovation opportunities; prioritising AI initiatives; and coordinating cross-team delivery of AI platforms and data pipelines. To boot, automation enabled by modern EA platforms cuts the time to the business value unlocked by AI by accelerating the generation of these insights.
It’s not just a case of the tools that an EA approach brings. The job role and purpose of EAs make them invaluable advisors in the AI transformation journey. EAs are well placed to carry out assessments of proposed AI projects, weighing an individual project’s feasibility, integration into the IT and business landscapes, and opportunities to leverage or re-use skills and resources from across the organisation. The journey that the EA discipline has been on means today’s practitioners are working in an agile, analysis-driven and democratic way. Their involvement in AI initiatives will supercharge such efforts, rather than stifle any progress, and cover critical aspects of AI transformation. EAs can, for example, coordinate robust data governance practices to ensure the security and compliance of AI solutions, which is a particular worry for many business leaders.
An EA approach pushes culture in the right direction
It’s important to stress that an EA approach to AI transformation goes a long way to bolstering an organisational culture that embraces AI-driven change.
Some reasons are obvious. Capability-based planning informs the strategic rollout of such change in a way sensitive to internal skill sets and what’s likely to yield business value that’s tangible to all teams. Some reasons are more subtle. When teams across the organisation are collaborating with EAs to feed data into AI initiatives, AI becomes less of a mysterious ‘black box’ for those teams. But that also requires EAs to open up their platforms and processes to a broader set of organisational users, so they also have visibility into the insights behind the roadmaps and can offer domain-specific expertise to further accelerate execution.
In conclusion, AI optimism needs to be tempered with AI realism. Business leaders need to focus on the right priorities and criteria to ensure success with the new technology. Greater chances of success beckon for leaders who see the opportunity and commit to an EA-driven approach to transition from AI pilot programmes to widespread adoption.
About the Author
Ed Granger is the Vice President of Product Innovation at Orbus Software where he’s responsible for executing product roadmap innovation. Ed is vastly experienced in the enterprise architecture (EA) space–previously serving as a Product Strategist at software vendor Ardoq as well as working as an EA practitioner in multiple capacities.
