Enterprise Knowledge Search
Oxide AI needed a more controllable and efficient way to turn large volumes of financial and market information into actionable investment insights without relying on opaque proprietary LLMs and fragmented cloud infrastructure.
Reported impact
Oxide AI improved model control and runtime performance while proving that watsonx.ai could support high-quality financial research outputs with less implementation complexity.
Oxide AI piloted IBM watsonx.ai to power an investment-intelligence workflow that blends quantitative and qualitative information. The company’s product looks beyond traditional stock screeners by combining market data with research papers, patents, news feeds and social signals, but that breadth created operational pressure when using proprietary models and scattered infrastructure. With IBM Client Engineering, Oxide AI used watsonx.ai as a unified environment for generative AI and machine learning so it could test more transparent, task-specific models and regain control over deployment. In the resulting design, large financial corpora and market signals are analyzed inside watsonx.ai to generate candidate insights, while the company evaluates quality and response-time performance against prior baselines. The pilot reduced dependence on external black-box models, simplified implementation and gave Oxide AI a clearer path to stable, explainable AI-driven research workflows.
Quantitative market data and qualitative sources such as patents, research papers, news feeds and social media are analyzed in watsonx.ai, where models generate research outputs that Oxide AI evaluates for quality and response performance.
In the proof-of-value (POV) development of its product, Oxide AI worked with IBM Client Engineering to pilot IBM watsonx.ai.
The solution achieved 37% faster average sequential response times than OpenAI GPT-4, while maintaining comparable quality levels.
Within less than a month of utilizing watsonx.ai, Oxide AI met a 95% qualitative acceptability threshold for the content it generated.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The official site footer lists "Oxide AB Västergatan 18B Malmö, SWEDEN 211 21."
Related
IBM Sample Jan 28, 2026 Bay Point Advisors Bay Point Advisors needed to speed private-market due diligence across secure data rooms full of documents without compromising answer quality, citation grounding, or document-level security controls. Hours to minutes Azure Blob Storage · Azure OpenAI · IBM DataStax Astra DB IBM Sample Jan 6, 2025 IRCEM Group IRCEM needed to answer growing volumes of customer questions about complex pension, insurance, and social-protection topics without forcing advisors to handle every inquiry manually by email or phone. 4,000 conversations per month IBM Cloud Expert Labs · IBM watsonx Assistant IBM Sample May 14, 2024 Sicredi Sicredi's support representatives had to answer questions across many financial products, and the previous assistant often required escalation to product specialists, which lengthened wait times for members and made… 1% improvement in net promoter score IBM Watson Discovery · IBM watsonx Assistant · IBM watsonx.ai 