Marvin OnumonuMultiverse’s artificial intelligence (AI) coach Atlas has shifted from a Socratic tutor to an intelligent guide for the full learning journey, now resolving 99.4% of learner messages and freeing up more time for human coaches.
The platform reported that Atlas assesses user intent and adapts its role, acting as a project partner, Socratic tutor, or support guide as needed.
Atlas works closely with human coaches for sensitive or complicated cases, which helps learners get faster resolutions.
Since launching its pilot in December 2023, Atlas has supported over 23,000 learners and processed 1.5 million messages.
The rollout has allowed Multiverse to scale upskilling and provide 24/7 support, while coaches have seen routine support queries halve from 41% to 18%.
Overall message volume per coach dropped by 32%, with more time now spent on complex learner needs.
Atlas resolved 9,461 messages in October, up from 1,160 in July when new support functions were introduced.
In October, 88.3% of all inbound queries were resolved by Atlas alone.
Less than 1% of messages escalated to a human coach, with the rest passed to a human support team.
Users gave Atlas a 99% helpfulness rating, even as it handled over 15,000 messages per day.
Additionally, Atlas resolved 99.4% of queries independently and maintained a 99% helpfulness rating from users.
Cem Gurkan, senior product manager at Multiverse, said: “We’ve moved Atlas beyond just Socratic coaching; it is now the first place a learner can go for any question they might have.
“It understands their intent and knows when it can help them directly, and when to connect them with human expertise.
“By handling the vast majority of all the support queries it receives, it gives our human coaches and our support team their most valuable resource back: time.”
Gurkan added: “They can now redirect their efforts from routine support to the high-impact, developmental coaching that actually transforms careers.”
Atlas uses retrieval augmented generation to query Multiverse documentation and support knowledgebases.
It also uses function calling to supplement responses with contextual information about the learner’s role, employer, programme, and skills needed.
Toni Sidall, a data learner from a global education nonprofit, said: “Using Atlas is like having an expert on call, giving straightforward answers and practical pointers.
“It made apprenticeship tasks feel simpler and more achievable – I used it for assessment clarifications, templates, step‑by‑step guidance and help on tasks when preparing evidence for my portfolio.”
