OneOffTech's Stories

Agent Mode is our way to embrace AI Agents browsing OneOffTech websites with markdown-first, AI-ready content, that is more accessible and visible.

Structured extraction with LLMs works best when input data is carefully prepared. We share how a parser-oriented approach — decoupling document parsing from AI tasks — enabled reliable extraction of lessons learned and recommendations from 200+ project reports.

Bridging knowledge management with controlled vocabularies and AI-powered retrieval. Working with M&E specialists, we built a SKOS-based knowledge graph, integrated three domain thesauri, and used it for RAG based retrieval.

Decoding the unstructured world. A week dedicated to document parsing, evaluation, and the open tools powering the next generation of data intelligence.

In traditional software, we debug behavior; in AI, we evaluate function. This post explores the tension between behavioral transparency and functional performance in AI systems, drawing on both philosophy and software engineering. When the internal workings are opaque—like in neural networks—we shift from analyzing how a system works to judging what it achieves.

Your AI system is only as good as your evaluation loop. A great model means nothing if you can’t measure and improve its real-world behavior. We introduce a practical approach for creating your AI-evaluation loop that's reliable, measureable and user-center, so you can add AI confidently to your product.

Does PDF parsing affects retrieval? Let's dive in our experiments on PDF parsing and the effect on document retrieval presented at the Berlin Buzzwords 2025 .

Over time, we've gathered a collection of links to useful services, libraries, and datasets. Now it's time to share this curated list of PDF tools with the world.

Need to test your application in a specific browser version? Here's a quick tip to help you get started using Playwright.

Agent is now a common term when referring to interactions with large language models. Let's explore some definitions following Anthropic guidance.

We explore the AI Act and its pillars from the perspective of companies/organizations using artificial intelligence.

The Retrieval-Augmented Generation (RAG) approach improves LLMs by incorporating domain-specific information for more accurate answers. Let's see how it works.

A workshop on knowledge management for all!

Explore how to handle personal and sensitive data when developing or interacting with machine learning applications. A case study focusing on Retrieval Augmented Generation (RAG).

Proof of Concept (POC) is a common tool used in technology-focused as well as knowledge management activities. We explore how the two souls compare.

OneOff-Tech is partner of the European funded DIGITOO project to improve digital literacy education of students. Autumn 2023 marks the start of the collaborative review and testing of the toolkit on media and digital literacy skills and the handbook on digital ecology.

OneOff-Tech is partner of the European funded DIGITOO project to improve digital literacy education of students. November 2022 marks the start of the journey to develop a toolkit for education professionals that includes information and pedagogical methods on digital citizenship, digital footprint, and digital ecology.

How OneOffTech is piloting Open Source solutions to power the communication and collaboration in the International Technical Higher Education Network (ITHEN).

OneOffTech is one of the contributors of the Multi-Donor Learning Partnership (MDLP) publication on how international development agencies are collaborating to deliver impact through knowledge, learning, research and evidence

Data-driven exploration builds value for knowledge discovery in expert networks. To illustrate, here is an example that comes from an assessment study for the Low Emission Development Strategies Global Partnership