BLX Learning Companion

The Learning Companion is embedded directly in the LXMS module as part of the learner dashboard and navigation, rather than appearing as an external support tool. Guidance is available where it is needed: within the current module, section and progress state.

Depending on the situation, the Learning Companion can display states such as Ready, Active, In Plan, Note, New Note or Critical. These are not decorative labels; they summarise signals from progress, failed attempts, periods without progress, momentum and coach recommendations into a clear assessment of the current situation.

Learners do not need to interpret a dashboard. They receive direct feedback on whether they are progressing as expected, whether guidance may help and whether a specific next step is recommended.

The Learning Companion turns signals into specific actions. Rather than merely flagging a problem, it provides a recommendation, a next step and, where useful, a brief explanation.

Triggers may include several failed attempts in quick succession, a prolonged period without visible progress, results below the expected level for the section, or a combination of errors followed by no recovery. The learner dashboard turns these signals into a concise summary, a recommendation, a clear next step and an understandable rationale. It can also offer direct follow-up actions, such as requesting further AI-assisted guidance or marking a suggestion as completed.

The recommended step can be linked to a specific content version, section, activity, language or audience variant rather than to an abstract module. This makes the reason for an intervention traceable and shows whether it remains relevant after a later revision.

Language is also part of the context for multilingual learning content. Guidance must be properly localised, using approved terminology, tone, examples and forms of address. AI can help formulate it, but does not replace linguistic and subject-matter approval.

The feedback remains specific and directly connected to the learner’s next action.

Many AI assistants used in education are disconnected from the learning design. They respond to individual questions without knowing the objective, assessment logic or planned learning journey. The BLX Learning Companion takes a different approach.

Its strength comes from access to the learning-design context. Objectives, audience, outline, storyboard, evidence, Bloom level and Cognitive Load Theory (CLT) guidance maintained in BLX Designer can all inform the companion.

This turns generic advice into feedback grounded in the learning design. The companion can consider the objective for the section, available evidence or assessment, whether review, transfer or simplification is appropriate, and which form of support suits the audience and module style. The result is relevant guidance rather than a generic motivational message.

This context matters when learning content changes over time. Guidance that was useful for version 1 may no longer be appropriate after a technical update, a new language version or a learning-design revision. The Learning Companion therefore needs both the learning history and the content version to which its intervention relates.

The BLX Learning Companion is designed to remain robust. Its rule-based operation provides meaningful support without AI, while an optional AI layer can add nuance, adapt the language and clarify the specific situation.

The model has two stages. A reliable rule base detects signals, thresholds and intervention patterns. Where the learning context and runtime data support it, an AI layer can refine the wording or make the guidance more specific.

Configurable parameters include thresholds for time without progress and failed attempts, the minimum interval between messages, whether coach guidance is enabled, an optional AI mode with rule-based fallback, and the tone and length of each message.

The Learning Companion remains controllable, and its rule-based support continues to operate if AI is unavailable.

This separation is particularly important in regulated or compliance-related learning. The rules define when an intervention is triggered; AI may explain, simplify or vary the wording, but remains bound by approved context, tone and limits. The Learning Companion is therefore a controlled support layer with optional language intelligence, not an unrestricted dialogue system.

Unlike static or heuristic help text, the BLX Learning Companion makes its guidance and the learner’s response measurable through xAPI and the LRS.

Recorded events can include displaying guidance, requesting an AI-assisted clarification, showing the response and marking a recommendation as completed.

Guidance becomes a measurable intervention. Its reach and use can be evaluated, providing a basis for systematic improvement.

These events should not be evaluated in isolation. They remain linked to the learning content, version, language, section and intervention pattern so that teams can distinguish between generally effective guidance, a problem in one version and a change in learner signals after a revision.

The BLX Learning Companion delivers its full value across the lifecycle of version-controlled learning content. It connects design, implementation, evaluation and revision. Guidance appears at the moment of need, while its effect can be assessed while the content remains in use.

Initially, the companion draws on the objectives, storyboard, assessments, audience and language. During delivery it recognises learning signals and provides context-sensitive guidance. xAPI and the LRS make use and impact visible, and the resulting evidence feeds back to authors, content owners and the learning architecture.

When learning content is updated, translated or republished, the companion’s context changes as well. Messages, thresholds, tone and intervention patterns can be linked to versions and releases, helping teams identify whether an improvement came from a better activity, clearer translation, changed interaction or adapted guidance.

Together with BLX Designer, this creates a closed loop: learning design provides the context for better interventions, while intervention data feeds back into the high-level concept, storyboard, analysis and blueprint.

The Learning Companion becomes part of the learning system throughout the digital learning lifecycle.

LXMS evaluation provides the wider data context: progress, time on task, repeated attempts, drop-outs, use of interactive formats and feedback. The Learning Companion turns this data into specific support. Consistently long completion times, failed attempts or abandonment within a section can prompt guidance for learners or an alert for authors and course leaders.

The two perspectives complement one another. The evaluation dashboard reveals patterns, while the Learning Companion determines whether they warrant an intervention during the learning process. Learner responses—whether guidance is viewed, clarified, accepted or ignored—are then evaluated in turn.

Evaluation moves from retrospective reporting to a feedback loop within the learning-content lifecycle. Teams can refine content, activities and media more precisely, while learners receive faster, more relevant support.

The LRS provides the wider context by combining companion events with progress, failed attempts, recovery patterns and drop-out signals.

The LRS dashboard shows where learners may be at risk, whether they recover after errors, how often guidance is displayed or clarified and whether recommended steps are taken. It can also compare rule-based messages with AI-assisted clarifications.

Patterns matter more than isolated cases. Repeated lack of progress, failed attempts or abandonment at the same point may indicate an unclear activity, insufficient preparation, excessive cognitive load, a poor translation or an unsuitable format.

The Learning Companion supports individuals in the moment while also helping teams identify systemic problems in the learning content.

The Learning Companion serves several audiences. Learners receive timely support: a brief assessment of their current state, a reliable recommendation, a meaningful next step and, where needed, an AI-assisted explanation.

For authors, aggregated guidance data becomes valuable when structural patterns emerge. Repeated messages or confirmations in one place may point to a misleading activity, an overloaded section, an abrupt rise in difficulty or a missing intermediate step.

For content managers, the data provides another source of evidence. It shows not only where learners need help, but which parts of the learning content need attention: a misleading activity, unclear translation, outdated example, sudden rise in difficulty or legacy content that lacks sufficient context.