Blog category
What AI actually changes in a company’s day-to-day work, and what is still an announcement. Topics here include language models inside existing workflows, permissions and data protection, the EU AI Act, and when an agent beats a hard-wired workflow. Dated assessments rather than forecasts.
7 articles
How a workforce of ten languages with no common tongue communicates seamlessly — real-time translation woven into the workflow, GDPR-compliant, tiered from on-premise to EU cloud.
Not "can the AI hallucinate?" but "what rights does it have?" OWASP Agentic Top 10, EchoLeak, Entra Agent ID — and what the mid-market must do about agent permissions now.
Is model size becoming less important than delivered intelligence? SLMs, inference-time compute, the LLMflation price collapse, and model routing — opinionated, with a fair counter-position on scaling.
Personalization across levels (segmentation to 1:1) and areas (onboarding, marketing, service, retention) — plus the GDPR guardrails and when it tips into "creepy".
The one hard deadline (2 Aug 2026), what GDPR, NIS2 and the AI Act now require of an automated workflow, and how to treat compliance as a build.
Practical use cases for AI in automation - no hype, real examples.
Stripped of the hype: what AI agents deliver in practice for teams of 10–250. With three setups we would never recommend.