MAST’s agents, and where each one stands.

Reads a tender pack and drafts an evidenced answer to every question in it, citing where each fact came from.

Drafting real tenders

Builds each person’s working day from the time-clock export and puts claimed hours beside tracked and calculated travel.

Scheduled weekday mornings

Drafts method statements and risk assessments in the office’s own template family, as an editable Word file plus a PDF.

Switched on; first real use not yet on record

Scans the government’s Find a Tender service for new opportunities every weekday morning and shows them in the HUB.

Running every weekday

Explains a six-month look-back of money in and out, by month, contract and site. The system computes the numbers; the agent explains them.

Built; switch-on still to come

Builds the teaching documents a teacher asks for, plan first: knowledge organisers, worksheets, revision guides and lesson slides.

6 jobs, 23 delivered versions, moved onto PRISM from the earlier portal

Answers technical presales questions in conversation, from a fixed knowledge snapshot.

32 of 32 pilot jobs reached done or a question for a person

Checks the certificates and delivery documents on a job against each other.

Switched on; first real use not yet on record

Reviews an expense claim against its receipts and the policy wording supplied with it.

Switched on; first real use not yet on record

Writes a message for each new order, plus a daily figures round-up, for the owner’s WhatsApp group. One agent per shop; enquiries too where that is switched on.

Live on the first shop; the second is switched on, awaiting its first order

Reviews the inbox twice a day and unsubscribes from mailing lists.

Running twice a day

Writes a morning briefing on new AI tools and techniques, judged against how MAST actually works.

Running every morning

Compares how each type of post performs against the brand’s own average and writes a do-more, do-less digest. It also scouts and ranks accounts worth engaging with.

Not yet on PRISM. Pulling live metrics and drafting replies are not built yet

Turns work chat messages into suggested tasks, matched to the right project.

Switched off; its source chats went quiet