All insights

Quarterly review19 March 20261 min read

From strategy to delivery

The UK government reported on a year of its AI action plan and started piloting AI assistants in public services, while India hosted the AI Impact Summit.

What happened this quarter

In January, the government published a progress update marking one year of the AI Opportunities Action Plan. It also announced a new cohort of AI fellows, funded through the Alan Turing Institute, to build open-source tools for public services, and a partnership with Anthropic to design and pilot an AI assistant for public services, starting with employment support. The Government Digital Service published guidance on making government datasets ready for AI.

In February, New Delhi hosted the AI Impact Summit, the latest in the series that began at Bletchley Park. The models kept improving, with Anthropic's Claude Opus 4.6 and OpenAI's GPT-5.3-Codex both released in February.

Our take

The tone in government has changed. Two years ago the questions were about principles and risk. Now they are about delivery: which services, which data, what evidence. That is a healthy shift, and it suits suppliers who can show working software rather than slides.

The guidance on AI-ready data deserves more attention than it will get. In our experience, the hardest part of most AI projects is not the model. It is finding the right documents, knowing which version is current and understanding who is allowed to see what. Organisations that sort out their data now will move faster later, whichever model they choose.

We also note the optional design of the employment-support assistant. Giving people a choice, with a clear route to a person, is the right default for public services.

Keep reading

Related insights