DSML: Combining Predictive and Generative Approaches for Innovative Data Science Solutions

In the midst of the rapidly evolving Generative AI (GenAI) landscape, it is crucial not to lose sight of the strategic decision-making aspects underlying AI platform development and deployment.

Last year, Gartner highlighted this important point in its Magic Quadrant report on Data Science and Machine Learning Platforms (DSML). While it hasn’t explicitly reiterated this focus in the latest edition, the analyst firm emphasizes that among the providers categorized as "Leaders," one has managed to find a well-balanced approach. That company is Dataiku, recognized for its ability to blend strategic vision with solid execution.

IBM Ascends to the Leader Quadrant, Replacing SAS

The latest Gartner evaluation features a total of seven vendors designated as "Leaders," including Altair, AWS, Databricks, DataRobot, Google, Microsoft—and newly promoted IBM. Previously classified as a "Challenger" in the 2024 report, IBM has made notable progress over the past year, particularly in the "Vision" axis of the Magic Quadrant, which assesses strategic orientation—covering industry-specific, geographical, commercial, marketing, and product strategies.

Conversely, SAS has slipped from the "Leaders" to the "Visionaries" quadrant. This shift reflects a decline in its "Execution" score, which gauges the platform’s ability to meet market demands through customer experience, pre-sales performance, and product/service quality.

The following overview details the current standing on the "Execution" axis:

Similarly, on the "Vision" axis, the ranking and movement are as follows:

Opportunities at the Intersection of Predictive and Generative AI

A year ago, Gartner emphasized the importance of platform complementarity. With data science activities becoming increasingly decentralized, relying on a single platform was deemed impractical. The American analyst highlighted the growing trend of building bridges between different vendors rather than choosing a one-size-fits-all solution.

This year, Gartner’s stance remains consistent, with particular attention on the evolving integration points between providers. The report notes a rising trend of targeting user profiles aligned with specific Business Units (BUs) and functional domains. Automated Machine Learning (AutoML) remains a fundamental element, now complemented by Generative AI assistants. These developments present unique opportunities for vendors to differentiate themselves based on how effectively they combine predictive analytics and generative capabilities.

On the functional front, the basic building blocks expected from leading platforms include:

• Importing or connecting tabular data
• Data preparation and cleansing
• Development environments supporting code-based workflows
• Model creation, evaluation, deployment, and lifecycle management
• Collaboration tools and project management features
• Administration of roles, permissions, and resource allocation

Optional features, such as data cataloging, data lineage, GPU support, MLOps integration, low-code development interfaces, and transparency-enhancing techniques, are often considered supplementary but add significant value depending on user needs.

Altair’s Acquisition by Siemens: What’s Next?

Last year, Gartner praised Altair for its flexible economic model based on consumable units across its product range. Its market understanding, evidenced by the success of deployments rather than platform usage alone, was also highlighted. The acquisition of RapidMiner, an important move into scientific applications, was seen as an opportunity to open new doors.

This year, Altair continues to stand out for its grasp of market trends, but more specifically in terms of its data fabric vision. Gartner emphasizes its Center of Excellence—offering a structured implementation program—and its recent acquisition of Cambridge Semantics, a leader in graph database technology.

In 2024, Gartner pointed out that the integration challenges stemming from Altair’s rapid acquisition of RapidMiner disrupted the coherence of its catalog. Additionally, Altair was considered less aligned with customer interest indicators compared to other leading vendors. Since then, Altair has been integrated into Siemens as of March 2025, raising questions about its strategic direction. Concerns are also raised regarding the firm’s relatively low branding around its AI fabric concept, which Gartner suggests needs more effective communication. The broad product portfolio, especially segments related to AI and analytics, could pose risks of confusion and integration challenges.

Challenges in Cost Management with AWS

Last year, Gartner appreciated the community support surrounding AWS’s DSML offering, along with the infrastructure services backing it and the security/privacy guarantees associated with foundational models available via Amazon Bedrock.

This year, focus shifts to the SageMaker Unified Studio. This integrated platform promotes collaboration and allows third-party providers to embed their tools without managing underlying infrastructure. Gartner commends the platform’s support for responsible AI features, notably Bedrock Guardrails and automated reasoning validation, which help verify the accuracy of generated responses.

In 2024, Gartner raised concerns about the limited recognition of Amazon’s proprietary "Titan" models and the potential complexity of AWS’s extensive foundational model portfolio, which could overwhelm users making platform choices. The report also pointed out that AWS seemed to prioritize machine learning engineering over more traditional decision-support uses.

In 2025, the challenges have expanded to include the entire foundation model catalog’s complexity, which may influence vendor selection. Cost management—particularly FinOps—remains a significant obstacle, affecting both infrastructure and SaaS services. Additionally, SageMaker Unified Studio still has room for improvement in integrating with other cloud platforms, limiting its flexibility in multi-cloud environments.

Databricks’ Steady Growth Amidst Platform Challenges

Previously, Gartner highlighted Databricks’ investments in GenAI and low-code development, its unified lakehouse + DSML approach, and the acquisition of MosaicML, which allows rapid and cost-efficient deployment of Large Language Models (LLMs).

This year, Databricks earns praise for the platform’s popularity among target audiences, enriching the skills market with a robust talent pool. Its stable leadership team and vision of a modular ecosystem—particularly relevant for finance and reputation management—are also noted as strengths.

However, there are challenges, especially concerning hybrid and modular data architectures. Gartner warned about potential performance management issues and the need for ongoing monitoring to ensure workload stability amidst rapid product evolution.

This year, additional emphasis is placed on the platform’s learning curve, which can be steep for new users. Furthermore, the intense competition around the integrated lakehouse + DSML approach and the absence of certain infrastructure features, like serverless GPUs, pose hurdles for adoption.

Dataiku’s Platform Usage and Market Visibility

Last year, Gartner recognized Dataiku for fostering excellent team collaboration, managing change effectively, and promoting a compelling vision for GenAI, exemplified by its LLM Mesh initiative supporting governance.

This year, Dataiku’s LLM Mesh continues to be well-regarded for governance enhancements. The company also benefits from consistent customer support, especially in initial use case deployment, maintaining its core focus on data science. The introduction of Dataiku Stories signifies an effort to expand into data storytelling.

However, its pricing model—combining platform fees with user licenses—was criticized in 2024 for its complexity. Community engagement and on-premises deployment are also less mature than those of key competitors, posing potential limitations. Moreover, reliance on functionalities provided by other vendors’ solutions often leads to overlapping capabilities.

DataRobot’s Position and Challenges in Data Science

Last year, DataRobot differentiated itself through ease of use and a keen market understanding, especially regarding multidisciplinary teams.

In 2024, it shifted focus toward building an application-centric ecosystem, notably through acquisition of Agnostiq and its orchestration platform Covalent, emphasizing operational deployment and enterprise applications.

However, it faces ongoing challenges: Gartner pointed out elevated pricing, concerns about strategic clarity—balancing AI platform ambitions against data management—and operational issues like high employee turnover and leadership changes.

The question persists regarding how effectively other software vendors are embracing DataRobot’s pivot to an agentic application ecosystem. Its offerings are perceived as less suitable for data scientists, with frequent interface updates making both code and no-code environments difficult to navigate.

Google’s Core Offerings and Usability Hurdles

Last year, Google received recognition for its foundational model catalog—including third-party models—along with balanced investments in GenAI and DSML, and rapid feature delivery.

This year, Gartner emphasizes the governance capabilities introduced through the integration of Dataplex into Vertex AI. The report highlights co-innovation efforts with clients and advancements in Retrieval-Augmented Generation (RAG) techniques, leveraging Google Search and enterprise data to manage structured and unstructured information.

Despite these strengths, Google lags behind competitors on governance and is more focused on ML use cases centered on data science. Its Vertex AI platform becomes less compelling for organizations that do not heavily use Google Cloud Platform (GCP), especially considering data storage and processing dependencies.

Gartner also points out that third-party support within Google’s DSML ecosystem is less mature, and the complexity of managing multiple RAG solutions could lead to confusion or inefficiencies.

IBM’s Broad Tool Suite Faces Awareness Challenges

IBM distinguishes itself with a broad array of tools—including frameworks, foundation models, and GPUs—and demonstrates innovation using models like Granite and implementations of AutoRAG and InstructLab. The recent acquisition of DataStax enhances its capabilities in Retrieval-Augmented Generation (RAG).

However, IBM faces notable awareness issues, especially compared to vendors offering more consumer-focused products. Its multi-cloud integration is uneven—in particular, less seamless with GCP than with AWS and Azure. The company also underplays the potential of its Software-as-a-Service (SaaS) solutions such as Watson for core data science activities, which may hinder wider adoption.

Microsoft Leverages OpenAI and Azure for Competitive Edge

Last year, Microsoft benefited from its extensive training resources and robust R&D, including developments in prompt engineering frameworks, large language models, and RAG architectures. Its Azure ML platform showcased a strong ecosystem of models, deployment options, and flexible pricing.

The year 2025 sees an ongoing focus on innovation, exemplified by Azure AI Foundry Labs, which offers experimental environments for AI frameworks. Its broad partner ecosystem and adaptable pricing models continue to be advantages.

However, prior shortcomings persisted: the data exploration capabilities within Azure ML were criticized, and frequent rebrandings of product lines added confusion. The frequent updates and the integration of GitHub Copilot necessitate ongoing adjustments for users, especially for data scientists.

Gartner notes that the performance gap between OpenAI’s models and those from other providers is narrowing, giving organizations more freedom to explore alternative solutions. The gradual rollout of Copilot for data science workflows remains limited, initially confined to notebook environments.

Through these insights, it is clear that the landscape of AI platforms is characterized by strategic balancing—between vision and execution, integration and differentiation, automation and human oversight. As companies navigate the complex terrain of generative and predictive AI, opportunities abound in how they optimize the confluence of these capabilities to meet diverse enterprise needs.