Skip to main content

From the blog

Thoughts on procurement, technology, and public money

Analysing the same bid with GLM-5.3 Flash and a tiered Claude stack
September 2026
Analysing the same bid with GLM-5.3 Flash and a tiered Claude stack
Europe relaxed its AI literacy rule. Procurement should raise its own bar.
August 2026
Europe relaxed its AI literacy rule. Procurement should raise its own bar.
China's open-weight models caught up on tender analysis
August 2026
China's open-weight models caught up on tender analysis
Roughly 86% of Latvian tenders are decided by price. The official figure says 51%.
August 2026
Roughly 86% of Latvian tenders are decided by price. The official figure says 51%.
We compared 11 procurement AI platforms on 26 features
August 2026
We compared 11 procurement AI platforms on 26 features
Software development day rates in all 27 EU countries
August 2026
Software development day rates in all 27 EU countries
Where public contracts must be advertised in the EU, and the four states where nothing forces it
August 2026
Where public contracts must be advertised in the EU, and the four states where nothing forces it
AI analysis of a EUR 68 million rail tender. It took 23 minutes.
July 2026
AI analysis of a EUR 68 million rail tender. It took 23 minutes.
What the Baltics buy: Estonia, Lithuania and Latvia procurement compared (2026)
June 2026
What the Baltics buy: Estonia, Lithuania and Latvia procurement compared (2026)
What a country buys: every tender Latvia published in the first half of 2026
June 2026
What a country buys: every tender Latvia published in the first half of 2026
What the 2026 CPO survey reveals about procurement, AI, and who really decides
June 2026
What the 2026 CPO survey reveals about procurement, AI, and who really decides
The rulebooks of public procurement: every major regulation and standard you should know
June 2026
The rulebooks of public procurement: every major regulation and standard you should know
Claude Opus 4.8 is now live for tender analysis. Here's what changed.
June 2026
Claude Opus 4.8 is now live for tender analysis. Here's what changed.
We ran Claude, GPT-5.5 and Gemini on the same tenders. Here's what we found.
May 2026
We ran Claude, GPT-5.5 and Gemini on the same tenders. Here's what we found.
Human-in-the-loop is not a weakness. It's the answer to "I don't trust AI."
April 2026
Human-in-the-loop is not a weakness. It's the answer to "I don't trust AI."