Retrieval Augmented Generation (RAG): Optimisation & evaluation techniques

Upgrade the RAG chatbot assistant based on user feedback as well as quantitative data from a broader evaluation framework. We discuss our implementation here.

November 21, 2025 · 18 min read

Foundation models for time series forecasting (2/2)

Time series foundation models enhance conformal prediction through higher forecasting accuracy and more stable calibration, especially in low-data scenarios.

October 17, 2025 · 11 min read

Foundation models for time series forecasting (1/2)

Foundation models offer a powerful alternative for time series forecasting, boosting accuracy and reliability across diverse applications.

October 16, 2025 · 9 min read

Rethinking “What If?” A Look at Semifactual Explanations in AI

Discover what semifactuals are, “even if” scenarios that expose AI’s strengths and limits while informing optimized decision-making.

September 8, 2025 · 18 min read

CLAIR.bot: One Year of Responsible AI Serving Public Debate

CLAIR.bot is a free, multi-agent AI that helps everyone navigate AI with sourced, clear answers. One year on, it keeps evolving to make AI debates more transparent and accessible.

July 24, 2025 · 4 min read

How to give sight to a blind language model

Get a good technical understanding of how vision-language models are built and trained.

July 23, 2025 · 15 min read

Under the Hood: Technical Blueprint of a GenAI-Powered ESG Due Diligence Tool

This paper introduces the Client ESG Due Diligence Tool, an automated solution that evaluates sustainability risks and opportunities of working with clients, whatever the industry they operate in.

June 10, 2025 · 25 min read

Revolutionizing ESG Due Diligence: A User-Centric Overview of a GenAI-Powered Sustainability Assessment Tool

This paper introduces the Client ESG Due Diligence Tool, an automated solution that evaluates sustainability risks and opportunities of working with clients, whatever the industry they operate in.

June 9, 2025 · 8 min read

Speed vs. Control vs. Flexibility: A Technical Framework for AI Architecture Selection

This article introduces a structured framework for evaluating AI Models, AI Workflows, and AI Agents, emphasizing empirical selection principles based on operational constraints and industry-specific requirements rather than technological trends.

June 6, 2025 · 6 min read