AI brings both benefits and challenges for enterprises; while some of its capabilities are overhyped, others are underestimated in terms of complexity. In this episode, Maryam Ashoori, Director of Product Management for Watsonx.ai at IBM, discusses the complexities of building enterprise AI applications such as the skills shortages and the rapid pace of innovation, and how IBM is focusing on addressing these challenges.

Ashoori talks about the potential of AI to drive productivity gains, highlighting recent statistics from IDC on the productivity increase from AI-driven coding systems. Ashoori also goes into the specific challenges of scaling AI in enterprise applications.

The current state of AI and its role in enterprise adoption

• Ashoori discusses the excitement surrounding generative AI, especially regarding the development of sophisticated applications. Ashoori points to the unique challenges of building on top of foundational models for reliable, scalable enterprise use.
• Ashoori emphasizes that developers are central to enterprise AI’s success, explaining that they face significant technical demands when working with complex AI stacks that differ from traditional tech solutions.

What are the complexities of building AI stacks?

• Ashoori goes into depth about the complex nature of building AI stacks, comparing them to simpler setups like the LAMP stack to highlight the added layers of complexity that AI brings to development.
• Ashoori explains the major integration challenges, such as aligning multiple AI frameworks and ensuring compatibility with existing models, templates, and guardrails critical for enterprise environments.
• Ashoori emphasizes the growing shortage of AI development expertise, pointing out that most enterprises lack the specialized AI skills needed to bridge gaps between traditional and AI-focused development.
• Rapid innovation within the AI field adds further complexity, as developers are pressured to keep up with evolving tools, which impacts adoption and creates friction for enterprise deployment.

How IBM is addressing skill gaps and simplifying AI development

• Ashoori explains the impact of recent layoffs in tech on the AI talent shortage, telling us that most enterprises primarily employ traditional app developers, highlighting the need to simplify AI stacks so non-specialists can build and implement AI solutions more easily.
• Ashoori discusses the importance of providing out-of-the-box toolkits and standardized templates that empower traditional developers to create AI applications without requiring advanced technical expertise.
• IBM is focusing on simplifying the AI stack and investing in training to close the skill gap, aiming to make AI more accessible across different levels of developer experience.

The role of automation and AI in boosting developer productivity

• Ashoori discusses AI’s impact on developer productivity, noting how AI introduces both complexities and efficiency gains. Ashoori talks about how it is reshaping the development process.
• Ashoori cites IDC statistics which show that developers expect AI coding systems to significantly boost productivity, validating the potential benefits AI offers to streamline coding tasks.
• Ashoori explains how AI automates repetitive tasks like code generation and test case creation, freeing developers to focus on more strategic areas that add greater value to enterprise projects.
• Ashoori shares an example of retrieval-augmented generation (RAG), explaining how AI can be used to automate RAG pipeline configuration, reducing developers’ workload on routine configuration.

The challenges of scaling AI in enterprise applications

• Ashoori talks about the specific challenges of scaling AI in enterprise contexts, and the transition many companies are making from AI exploration to operational use. Ashoori emphasizes the challenges presented by large models in terms of cost, latency, and environmental impact.
• Ashoori discusses the industry trend toward smaller, customized AI models designed for specific data and business needs, allowing enterprises to achieve higher efficiency in AI deployment.
• Optimizing AI solutions for ROI is essential, and companies need to focus on cost-effective and resource-efficient AI approaches to maximize long-term enterprise value.

Guest: Maryam Ashoori (LinkedIn)
Organization: IBM (Twitter)
Show: Let’s Talk

This summary was written by Emily Nicholls.