This from Mustafa Suleyman - CEO of Microsoft AI - is deeply dishonest.

Sounds like Barnsley will be the first town run entirely by vendor presales.

oh please, that's a ridiculous comparison.

Finally, this product naming is a utter mess and is going to end up causing everyone headaches trying to track products, revenues etc in the future. Spraying the Agentforce brand on everything - old and new - is beyond dumb.

2 final things here of note; this is interesting commentary from Benioff on the earning call. This is something we wrote about in 2024, but it's hard to know whether this is just part of Salesforce's messaging focus on LLM shortcomings, or they're agreeing with our thesis from back then.

Well that's excellent work there Google, thanks.

Can someone with access to UK Housing Secretary Steve Reed hold some sort of intervention?

Anatomy of an AI story;

Headline: well that sounds cool
Paragraph 5: oh, well I'm sure the tool helped too, right?
Paragraph 14: no, it's a pre-emptive tool, so couldn't have helped in this case at all, wtf?

Peter Kyle - UK Secretary of State for Business and Trade - seems to have zero clue that "exploring research and knowledge" is the very root of the eventual high-tech entrepreneurism he wishes to encourage and does a massive disservice to students in doing so.

A month on and as predicated, this data - here in a report about a UK gov deal with Google - is used as indicative of the potential for the technology.

I can't help feeling that this statement from Anthropic - who let's not forget were the developers of MCP in the first place - rather diminishes is as a USB for Agents (which, let's face it was a stupid way of describing it in the first place).

Us last month / The Reg this morning

Finally, it's the "users reported an average of 26 minutes saved per day" that's likely to generate the most repetition; I'm not suggesting that it indicative in the data. It's just when set against the other data in the report, it doesn't carry a huge amount of practical meaning.

The detailed data on use frequency was user reported, rather than through interaction data; the report claims "respondents used M365 Copilot heavily throughout the trial", though - as far as can be ascertained - daily use for Excel was ~15%, Word ~25%.

I'm interested especially in the adoption data - described as "strong" in some of the press coverage, which is defined in the report (not in the methodology for some reason, but with the results section.... but I'll let that go).

2024's AI Assistants chasing you from tab to tab like...

The vendor case study can be found here (summary in image); www.behold.ai/case-studies...

Hey UK Gov, if you want to remain credible on AI maybe don't feed these highly misleading bits of data to political journalists (as they apparently won't fact check them). It serves nobody well.