Engineering teams that need a past qualification temperature or a subsystem test result usually have to dig through documents, systems and colleagues to find it. Montreal-based Maya HTT has launched M-Bot, an engineering knowledge assistant that answers those questions from internal records and links each answer back to its source.
M-Bot runs on-premises or within customer-controlled infrastructure. Maya HTT is aiming it at industries such as aerospace, defence and manufacturing, which handle sensitive intellectual property and face data sovereignty requirements.
The assistant uses a retrieval-augmented generation architecture tuned for engineering and manufacturing content. Maya HTT says it handles text, images, tables, graphs and other technical artefacts that conventional large language models cannot parse.
Users ask questions in natural language. Examples include historical qualification temperatures, previous technical issues, manufacturing processes and subsystem test results, with answers linked to supporting source material.
M-Bot connects to Teamcenter and other product lifecycle management environments, as well as enterprise repositories such as SharePoint and Confluence, with full source traceability.
Maya HTT claims M-Bot retrieves information up to 60 times faster than a manual search. It says a five-minute manual search is reduced to about five seconds.
Pricing is fixed rather than usage-based. M-Bot carries no large language model token metering, so costs are known up front.
"Engineering organisations already possess enormous amounts of valuable technical knowledge, but we've found that it's challenging to get the right information to the right person in time to influence key decisions," said Remi Duquette, VP Industrial AI at Maya HTT.
"We developed M-Bot so teams could access that knowledge in seconds and keep sensitive engineering data completely under their control," he said.
Maya HTT is an engineering software and technology company working across simulation, software development, AI and digital engineering.
M-Bot is available now for demonstration and enterprise deployment.