Logo

Turning AI and automation into measurable impact for financial services.

The first 90% of the code accounts for the first 90% of the development time. The remaining 10% of the code accounts for the other 90% of the development time.

by The Ninety-Ninety Rule (Tom Cargill, Bell Labs) 1985

Scroll Down

Services

AI & Automation Strategy

Advisory on the practical application of AI and automation within financial services, with a strong focus on governance, portfolio roadmaps, and business value. This includes identifying where techniques such as Retrieval-Augmented Generation (RAG), agentic toolchains, and workflow automation create meaningful impact—and where they do not.

Architecture, RAG & Toolchains

Design and assessment of modern AI architectures, including RAG-based systems, AI toolchains, and agent workflows. Coverage includes model selection (open-weight and proprietary), quantization strategies, inference approaches (on-premise, hybrid, inference providers), and secure integration with existing data, APIs, and software architectures.

Implementation & Delivery

End-to-end leadership of AI initiatives from concept to production. Implementation can be delivered through existing in-house teams (preferred, with structured upskilling), executed in collaboration with trusted third-party providers, or carried out directly where speed, confidentiality, or complexity requires it.

Upskilling & Adoption

Targeted enablement of technical and business teams to ensure sustainable ownership of AI solutions. This includes hands-on knowledge transfer around RAG, toolchains, model operations, and governance, combined with executive alignment to anchor adoption.

Governance & Value Measurement

Definition of governance frameworks, policies, and portfolio-level KPIs to ensure AI and automation initiatives remain compliant, scalable, and focused on measurable business and operational outcomes.

About

Urs Gubser

Urs Gubser

In recent years, the focus has been on designing and implementing AI-based solutions in real-world, regulated environments. This includes hands-on work with Retrieval-Augmented Generation (RAG), agentic toolchains, and AI-enabled workflows, using a broad spectrum of models—open-weight and proprietary, quantized for efficiency, and deployed via on-premise, hybrid, or inference-provider-based architectures. The work goes beyond experimentation, emphasizing production-grade AI systems that integrate with existing data, APIs, and software architectures, align with governance and compliance requirements, and deliver measurable operational and strategic impact. Experience covers the full lifecycle, from architecture and model selection through implementation, adoption, and continuous evolution.

Please contact for further information and engagement opportunities.

Urs Gubser has held roles across both technology and business, ranging from software engineer and system administrator to business analyst, project manager, program manager, product manager, and innovation and transformation leader. His experience spans global financial institutions and technology providers including UBS, Deutsche Bank, Banc of America, Citi, Millennium Partners, SIX Payment Services, and Worldline.

Contact Us

Public Projects

The following selection of projects highlight the use of LLMs, vibe coding, and clustering applications within products. These are all vibe coded and run on Google's Firebase.

Screenshot placeholder for the AI playground project

Developing a SaaS based AI Playground

Illustrates how open source models can be utilized in corporate settings because they can be deployed internally.

Screenshot placeholder for the meeting time evaluation project

Using LLMs for date and meeting time evaluation

Accepts free-form availability inputs for large groups and suggests meeting slots without manual polling.

Screenshot placeholder for the news clustering project

Clustering & summation

Crawls news sources, cleanses copy with GenAI, clusters related stories, and pairs summaries with generated imagery.

Blog

Newest Posts

Jun 13, 20263 min read

Opus 4.8 vs. GPT-5.5: The 3D Maze Runner Rematch

A Head-to-Head Comparison of AI Coding — Round Two Introduction A while back I pitted Opus 4.5 against Kimi 2.5 in a 3D maze runner build-off. This...

Jan 31, 20265 min read

Opus 4.5 vs. Kimi 2.5: The 3D Maze Runner Showdown

A Head-to-Head Comparison of AI Coding Introduction Over the weekend, I put two leading AI models to the test: Opus 4.5 (using Claude Code) and Kimi...

Nov 30, 20256 min read

MCP + Context: engineering for the context – hard lessons learned

Intro I have built my own orchestration framework because most of what I’ve seen was too complex or tried to lock you into creating workflows a...