# Multiverse says its Compass AI reviews every instructor session > Multiverse says its Compass AI reviews instructors' sessions for quality and suggests development areas. Teachers told The Guardian that the system feels like constant scrutiny; the company says human managers make employment decisions. By BIG CHANGE Editorial Published: 2026-09-28T13:47:31.968Z Updated: 2026-09-28T13:47:31.968Z Canonical: https://bigchange.ai/blog/multiverse-compass-ai-instructor-session-review ![Conceptual charcoal illustration of an instructor gesturing beside a plain whiteboard as adult learners sit facing the lesson in a training room.](https://bigchange.ai/api/media/file/hero-v1.png) AI-generated conceptual illustration by BIG CHANGE. Multiverse has begun using an AI tool called Compass to review every session its instructors teach. The company says it gives managers a way to spot possible problems and offer more specific coaching. Instructors who spoke to [The Guardian](https://www.theguardian.com/technology/2026/sep/28/teachers-euan-blair-firm-report-horrendous-stress-ai-used-to-rate-their-work) described a different experience: feeling watched throughout their work and anxious about what the system might flag. The accounts describe the same move from occasional human observation to routine AI review. They do not establish that Compass makes employment decisions. Multiverse says those decisions remain with people. ## The big change - **What changed:** Multiverse says a manager previously observed an instructor about once a month. Compass now reviews every teaching session, turning individual lessons into a continuing source of feedback and possible risk flags. - **Why it matters:** A manager can check a flagged session sooner, but the wider view also changes the instructor's working conditions. Instructors told The Guardian that scrutiny and the possibility of being questioned over a lesson have caused stress. That is reported testimony, not an independently measured effect across the workforce. - **What to watch:** Multiverse says instructors will see the same reports as managers in October. That promised access will make the company's distinction between AI suggestions, human checks and performance management more visible to the people being assessed. ## What Compass reviews In a [September 25 post](https://multiverse.study/blog/how-were-using-ai-to-invest-in-our-coaches), Multiverse says Compass compares instructors' sessions with its Compass Framework and combines its findings with learner feedback. It produces a review for each session, showing managers where quality may be at risk, and suggests development areas drawn from an instructor's teaching over time. The company describes the output as a prompt for human review, not a verdict on an instructor. The Guardian reports that transcripts of online classes are analyzed and scored. Its account says the system can flag issues such as an instructor failing to resolve a connection disruption quickly, relying heavily on filler phrases, or giving unclear task instructions. Those examples come from the newspaper's reporting on the system; Multiverse's public post does not publish the full Compass Framework or the instructions used to generate each flag. Multiverse says delivery managers check Compass reports against what they know, decide whether to act and agree development goals with instructors. It says performance management rests on human observation and Compass makes no decision about anyone's job. A spokesperson told The Guardian that the most consequential action the agent can take is to recommend that a human review a session. Those statements describe the company's policy and intended workflow; they do not, by themselves, show how every flag is handled in practice. ## What instructors reported Several instructors spoke to The Guardian on condition of anonymity. One said the stress since the system's introduction had affected sleep. Another said the need to avoid an AI mark was diverting attention from delivering the lesson and listening to learners. The newspaper also published an instructor's first-person account saying managers could investigate apparent departures from the teaching script that were actually responses to learners' needs. These are firsthand accounts as reported by The Guardian. BIG CHANGE has not interviewed the instructors or measured the system's effect on staff wellbeing. Their testimony does, however, identify a practical question for any continuous review system: whether a flagged passage captures poor teaching or loses the context that made the instructor's choice appropriate. The rollout also limits what instructors can currently check for themselves. Multiverse says managers alone saw the reports during August testing and September introduction. It says instructors will gain access through its internal coaching platform in October. The Guardian's first-person contributor likewise said managers could see the dashboard while instructors could not at the time of writing. ## Where the decision sits There are several steps between a session and a job consequence: Compass generates a review, a manager checks a possible concern, and a person decides what action to take. The company says formal performance management depends on human observation. The Guardian's reporting shows why that distinction matters even when no automated employment decision is made: a flag can still trigger a manager's scrutiny and shape the feedback an instructor receives. Multiverse's [published AI principles](https://multiverse.study/blog/introducing-our-ai-principles) call for transparency and human guidance. The company's October access promise is a concrete test of those commitments for instructors, although access alone would not establish that the assessments are accurate or that staff can challenge them. The available public material does not tell us the system's error rate, how often managers reject flags, or whether any individual employment outcome has depended on a Compass report. BIG CHANGE did not test Compass. The issue here is employers using AI to assess teaching work. It differs from [universities' rules for AI-assisted marking of students](https://bigchange.ai/blog/australian-university-ai-marking-rules) and from [the independent human judgment sought in AI-training work](https://bigchange.ai/blog/ai-trainers-human-judgment-independent-evaluation). In each case, the useful boundary is specific to the work being judged and the authority assigned to the system. ## Sources - [The Guardian: Teachers at Euan Blair firm report ‘horrendous stress’ after AI used to rate their work](https://www.theguardian.com/technology/2026/sep/28/teachers-euan-blair-firm-report-horrendous-stress-ai-used-to-rate-their-work) — Robert Booth's original reporting describes transcript review, system flags and anonymous instructors' experiences; its first-person contributor describes feeling scrutinized and disputes some flagged interpretations. Those accounts are attributed testimony and reporting. BIG CHANGE did not interview the staff or obtain underlying Compass records. - [Multiverse: How we're using AI to invest in coach development](https://multiverse.study/blog/how-were-using-ai-to-invest-in-our-coaches) — Company description of Compass: every-session review against its Framework, learner feedback, session-level risk signals, suggested development areas, human manager checks, and a stated boundary around employment decisions. It promises instructor access to reports in October; the post does not establish accuracy or case-level practice. - [Multiverse: Introducing our AI principles](https://multiverse.study/blog/introducing-our-ai-principles) — Published commitments to transparency and human guidance. The principles state company intent and do not validate Compass or demonstrate that its assessments can be challenged. - [BIG CHANGE: Australian universities set different rules for AI in marking](https://bigchange.ai/blog/australian-university-ai-marking-rules) — Related coverage #112 concerns AI-assisted evaluation of students. This article examines AI review of instructors, a distinct use of automated assessment in education. - [BIG CHANGE: When an AI trainer uses AI, the missing product is human judgment](https://bigchange.ai/blog/ai-trainers-human-judgment-independent-evaluation) — Related coverage #25 concerns independent human judgment in AI-training contracts, a different evaluation problem. It is not evidence for the Multiverse case.