Scaling gen AI: 13 elements for sustainable growth and value

Deloitte has identified 13 elements across strategy, process, talent, and data and technology to help organizations take the next step to sustainably scale generative AI for strategic business value.

Near the top of every enterprise agenda is a question of how to leverage generative AI. With use cases proliferating horizontally across functions and vertically within business units, the next step is...sustainably scaling gen AI for strategic business value.

Getting more gen AI into production

Deloitte’s State of GenAI in the Enterprise report revealed that many businesses are encountering challenges when making the transition from gen AI proof-of-concept to scaled deployment.

  • Nearly 70% of respondents said their organizations have moved 30% or fewer of their gen AI experiments into production.
  • Forty-one percent of organizations struggle to define and measure the impact of gen AI.
  • More than one-half of organizations reported they are avoiding certain gen AI use cases because of data-related issues.

This suggests that while enterprises are investing in gen AI, they are not yet seeing the full potential return on investment. Another common challenge is defining what is required to achieve gen AI scale at a practical level.

How do we define AI at scale?

We define scale broadly as the ability of a system to handle a growing amount of work or its potential to be enlarged to accommodate growth with steadily decreasing unit costs. For gen AI specifically, scaling also means moving from experimentation to implementation in a way that is sustainable, secure, and aligned with business goals.

Gen AI at scale generates more diverse and representative outputs, it can handle more complex tasks, and its speed, output quality, and accuracy are enhanced. As a result, operational costs become more efficient and business impact is governed, measured, and communicated.

Essential elements for scaling AI

At the highest level, gen AI scaling factors can be grouped into the familiar areas of strategy, process, talent, and data and technology, with 13 essential elements represented in the honeycomb below.

While each area presents challenges to be navigated, let’s look at the leading practices that help point the way to gen AI value realization.

  • Strategy
    1. Ambitious strategy and value management focus: Establish a comprehensive vision with a top-down mandate.
    2. Clear, high-impact use case portfolio: Explore low-barrier, high-impact use cases to drive efficiencies and savings.
    3. Strong ecosystem collaboration: Evolve with existing providers and new market entrants alike.
  • Process
    1. Robust governance: Create repeatable governance processes to help standardize work.
    2. Integrated risk management: Address risk and data security across the gen AI lifecycle.
    3. Agile operating model and delivery methods: Support a cohesive approach to orchestrating the components.
  • Talent
    1. Transparency to build trust in secure AI: Help stakeholders understand the gen AI vision and how it creates value for them.
    2. Transformed roles, work, and culture: Nurture adoption by documenting responsibilities and process amendments.
    3. Acquiring (external) and developing (internal) talent: Balance talent acquisition with workforce upskilling.
  • Data and technology
    1. Modular architecture and common platforms: Prioritize a flexible IT architecture to facilitate enhancements.
    2. Provisioning the right AI infrastructure: Take an agile approach to enable continuous improvement.
    3. Modern data foundation: Align data capabilities and processes with gen AI strategy to support quality and accessibility.
    4. Effective model management and operations: Monitor for impartial output accuracy and focus on cost management.

Measuring success with gen AI at scale

The value of scaled gen AI deployments is found in how they advance an integrated enterprise strategy and drive toward business goals. Establishing realistic goals for quantitative key performance indicators (beyond productivity and efficiency metrics, such as hours saved) allows the enterprise to assess whether the scaled deployment is achieving its intended business impact. With a use case portfolio that balances cost- and revenue-oriented value levers, there are key indicators that reveal whether the enterprise is on the right track:

  • Increased speed to market, from ideation to deployment
  • A decline in proof-of-concept demand, as demand shifts to low-code environments available to business users
  • A decrease in unit cost for new capabilities/solutions, with technical solutions and code being reusable, thus reducing development efforts
  • An increase in the number of foundational capabilities that help the organization access gen AI advancements as they emerge
  • An increase in domain-specific models allowing for more use cases and broader application across the organization
  • Increased use of capabilities and solutions, owing to a growing number of users in the enterprise
  • An increase in stated value realization on a cumulative basis due to gen AI
  • An increase in internal certification/badging of existing employees in gen AI capabilities, both functional and technical
  • Use of gen AI to redefine a business process, rather than embedding gen AI in existing business processes

An evolving approach to gen AI strategy

Gen AI capabilities are improving and multiplying, and at this point, few organizations are likely to have achieved each element of AI scaling to their greatest capacity. The leading practices, processes, and ecosystem of complementary technologies are still being developed and defined.

While change is inevitable, pursuing the elements of scale today positions the organization to go live with gen AI for business value as this transformative technology evolves.

Note: This article was created by Deloitte.