Project management edition
Meta’s Muse Wonder scores well on reporting, but sometimes misses details in transactions.
Muse Wonder is strong at reporting, but not at every accounting entry
Meta’s new model is strong at accounting questions, but makes mistakes with everyday transactions.
Meta released Muse Wonder on 12 October. Meta calls it its strongest model yet. It came second in an accounting test with 57 AI models. It completed 83.2 per cent of the tasks correctly. Its predecessor, Muse Spark 1.1, scored 61.4 per cent. Wonder was especially good at reporting. This included questions about cash flow, the money coming in and going out, and financial statements. It made more mistakes with everyday transactions. It made errors on some supplier payments. Results also varied for entries made at the end of the month.
For project work, a good report is not the same as a correct individual entry. A clear summary of budget data does not prove that every transaction has been recorded correctly. The source describes a benchmark. This is a set test used to compare models. It was not a trial using project accounts. So the figures do not predict how much benefit the model will bring to your project.
The test gave the models accounting tasks and compared their answers. Wonder did well on reporting questions. Most errors were with everyday transactions and exceptions. So a model can be strong at one task in a field, and weaker at another. Meta did not release a safety report when it launched the model. The score comes from an accounting test. It does not show how the model performs in every project.
Background for subscribersMuse Wonder is good at accounting reports. But it also makes mistakes with everyday transactions and month-end entries. The test does not show how well the model will work in your project. So check important entries yourself, even if the summary looks good.
Copilot can schedule meetings from chat
Microsoft describes a feature that lets Copilot schedule appointments and meetings from chat. The assistant finds free times, books rooms, creates an agenda and sends invitations. The feature appears on the roadmap dated 15 October. This is the list where Microsoft announces plans for new features. The rollout is for US government environments, such as GCC, GCC High and DoD. The source does not say that organisations in the Netherlands can already use the feature widely. You ask Copilot to schedule an appointment or meeting. Copilot finds free times, books a room and creates an invitation with an agenda. So a chat about plans can become a booking straight away. You get more than just a suggestion in text. The roadmap lists only US government environments. It does not say that organisations in the Netherlands can already use the feature. The source also gives no price or licence terms.
Background for subscribersThis is a taster.
Subscribers read the whole Project management edition: every story, each with its background in plain language.
Subscribe →Not ready for that? Read the free letter first →
AI and tech, made simple.
This is the Project management edition. Subscribers get the background to every story, in plain language.
Subscribe
inklaretaal · Amsterdam · © 2026