Arria NLG, which is listed on the Alternative Investment Market, said its Natural Language Generation (NLG) products are generating predictions for around 5000 locations across the UK on the Met Invent portal.
NLG is a branch of computer science that helps to process large sets of data into readable text and graphics for human consumption.
Arria's product range comes from research done by Professor Ehud Reiter and the Data2Text Aberdeen University spin-out he co-founded. Dr Reiter is now also chief scientist at Arria.
While Arria has been working with the Met Office since 2009 the latest update is the first time NLG generated narratives have been used for five-day forecasts.
The company, which is based in London but still has staff in Aberdeen following its acquisition of Data2Text, said it hopes NLG weather forecasts could be used for millions of online sites around the world.
Arria said: "This greatly increased level of text forecasting granularity would be virtually impossible using human-authored forecasts."
John Harrison, Head of Products of the Met Office, said: "Natural Language Generation is an opportunity for the Met Office to reach users with site-specific forecasts incorporating clear and simple narrative guidance on the weather in their immediate locale."
Stuart Rogers, chairman and chief executive of Arria, said: "We are very pleased to be working with the UK's national weather service and its science team to write NLG-authored on-demand, site-specific weather forecasts. Natural Language Generation automates expert analysis and reporting not only in meteorology but in a variety of fields to make optimal use of Big Data."
Shell is also using Arria products for instrument monitoring in the Gulf of Mexico.