The rise in SaaS surprised many CIOs and their teams, quickly overwhelming IT through dozens of rogue ‘Shadow IT’ applications. For a short while, IT’s full-time agenda was regaining app control, mitigating data use risks, and reestablishing compliance and customer protections.
GenAI’s rapid rise and organization-wide experimentation are bringing on a heavy dose of déjà vu. Unfortunately for IT, ‘Shadow AI’ risks and concerns eclipse those of Shadow IT.
SaaS applications are about productivity. Most are flavors of products (e.g., CRM, HR, ERP) that IT already understands and provides. This made it relatively easy to align security and governance updates with corporate policies and train users with online courses over a few hours. SaaS also has inherent native security, with either the vendor responsible for securing where the data resides or the application acting as a layer on the organization’s existing infrastructure.
Shadow AI is an altogether different beast. It’s unclear with GenAI how data is used, leveraged or stored by models. What is clear is that data is out of the company’s walled garden. Models aren’t transparent, and even creators admit it’s unclear exactly what happens on these platforms. This leaves enterprises skeptical about how proprietary data is/will be used, and how much of their copyrighted data is already in use without permission.
These issues are driving organizations to explore private language models. While seemingly more secure, this approach has its own challenges. Internally hosted models require constant monitoring and tuning with escalated compute and storage costs. IT teams usually don’t have all the requisite GenAI skills, and only the largest organizations have budgets that match GenAI’s needs.
Despite all this, GenAI’s interest and adoption marches full steam ahead. The Board and C-Level are pushing hard, driven by promises of huge productivity gains, while employees are driven by fear of being left behind and automated out of their jobs.
As an industry, we must address these conflicting priorities with a guided and governed approach that accounts for what’s unique and different with GenAI.
The GenAI Center of Excellence (COE)
Like a traditional COE, a GenAI COE aligns stakeholders to ensure organization-wide, secure and reliable technology deployment and usage. However, unlike rolling out a new ERP system with a stock two-day training, GenAI success at scale won’t be achieved with a one-size-fits-all approach. No individual model, cloud or training session can deliver on the universe of potential use cases and individual/BU needs that GenAI creates. Organizations need an agile and flexible approach to handling rapid technology shifts, evolving business requirements, and new data demands to realize GenAI’s potential.
A GenAI COE considers how each use case relates to the larger strategy, with chapters and unique goals depending on where and how the technology is deployed. There is an equal focus on technology and change management, creating the business rules, technology competency, program management, innovations, and partnerships to deliver AI and GenAI capabilities into every process, product and experience. There is specific focus on:
- Identifying user journeys and use cases. IT & departments partner to understand BU needs/maturity. This informs trusted data foundation design to provide the right data and model to the right roles within the context of an overall journey.
- Clear architecture and technology set blueprints that balance cost, resilience and reliability. There are limited funds for GenAI, so GenAI projects must enhance, enrich and overhaul existing operations where applicable to pay for itself.
- An operating model and execution playbook to support innovation with security and safety, so rollouts meet the business where it is and enable stakeholders to participate in mutual success.
- Individualized GenAI skills training and literacy as part of ongoing change management to ensure safe and appropriate adoption.
- Constantly evaluating the build-buy-invest-partner balance so the organization can effectively embrace and benefit from innovations.
- Business value identification and measurement. The emphasis must be on deploying technology to enhance performance and profit, not to just be “cool.”
GenAI and AI have incredible potential to bring data together in new ways, delivering unique experiences to every employee and customer with augmented performance and improved bottom-line value. A dedicated GenAI COE is crucial to companies confidently organizing, planning, and embedding these technologies while also getting ready to democratize GenAI across the organization.
