AI OBSERVABILITY
The rise of AI as the fuel of corporate growth also means a need to manage increasingly complex systems and data pipelines. This calls for providing managers and professionals with “the ability to see and understand the state of the system,” said Baris Gultekin, head of AI for Snowflake. This not only “examines the performance of the system itself, but the quality of the outputs of a large language model—including accuracy, ethical and bias issues, and security problems such as data leakage.”
The coming year “is the year AI observability goes mainstream,” he said. “It’s the missing puzzle piece to building explainability into the development process, giving enterprises faith in their AI demos to get them across the finish line.”
The rising demand for AI observability means new solutions will be flooding the market. “However, while there will be many AI observability startups, observability will ultimately end up in the hands of data platforms and the large cloud providers,” Gultekin stated. “It’s hard to do observability as a standalone startup, and companies that adopt AI models are going to need AI observability solutions, so big cloud providers will be adding the capability.”
AUTOMATED AI
AI may be the ultimate form of automation, but it also demands automation to build and deliver it. “Simply put, automation is needed to solve AI’s complexity,” said Bill Lobig, VP of product management for IBM Automation. “Organizations will advance and scale their AI initiatives using automation, moving from spending time managing and maintaining AI applications and IT environments to proactively detecting and resolving issues.”
Next year, Lobig predicted, “You won’t be able to have an AI conversation without talking about automation, and vice versa—you cannot have an automation discussion without talking about AI.”
COMMODITIZED LLMS
Large language models (LLMs) burst on the scene in recent years, led by the “frontier” and models such as OpenAI GPT, Google Gemini, and Anthropic Claude. These publicly available LLMs serve a broad range of needs, but now, there is a need for more specialized models.
“In 2025, LLMs will become commoditized, leading to AI pricing models collapsing as base-level capabilities are offered for free,” said Udo Sglavo, VP of applied AI and modeling for R&D at SAS.
“The real value will shift to specialized services and domain-specific applications built on top of these models,” said Sglavo. “Simultaneously, the rise of open source LLMs will challenge the dominance of a few key providers, driving a more decentralized AI landscape where customization and integration will be the key differentiators.”
SEEKING DATA ROI
There has been a tremendous surge in activity and purchasing of next-generation technologies, including cloud adoption, app modernization, DevOps, Agile frameworks, and serverless data analytics.
The problem is, “Many companies are getting to a point where they are evaluating if they are seeing all the tremendous ROI benefits they were promised,” said Jaret Chiles, chief services officer for DoiT International.
“I believe we’ll see more companies re-evaluate their application deployment strategies,” said Chiles. “The questions they will examine may include: ‘Did the well-optimized, heavily micro-serviced deployment designed for near limitless horizontal scale turn out to save money?’ ‘Did the product end up not needing the level of scale it was designed for? Is there a lower cost way to deploy the same application?’ ‘Is the team leveraging tons of costly deployment tools that are rarely used, or difficult to justify against a deployment schedule that has stagnated as the customer base stalled?’ ‘How optimized are your fully loaded costs really?’”
There are so many tools “to leverage and amazingly powerful new technologies available, but with maturity comes time to make smarter decisions about what actually makes sense for your unique business and your use case-specific application strategy,” said Chiles.
In the year ahead, the great re-evaluation will begin. While investments will continue to pour into next-generation technologies, decision makers will also be stepping back to evaluate what is working for their organizations.