AI Features, Priced Separately

Across the LMS market in 2026, AI is being unbundled. Vendors are putting AI capabilities into a separate tier or surcharge so the line is optional, removable at renewal, and visible as a value-prop on the order form.

• AI-assisted authoring. Draft generation, image and asset generation, tone and reading-level adjustment. Reduces SME bottleneck on content creation.
• Automatic tagging and metadata. AI extracts topics, skills, and learning objectives from existing content; powers search and recommendations.
• Playlist or learning-path curation. AI assembles personalised paths from existing content for a learner, role, or goal.
• AI assessments. Question generation from content, AI grading of short answers, scenario-based assessment generation.
• AI assistant or chatbot. In-product Q&A, guidance, content lookup, and increasingly action-taking (assign me this course, find me a course on X).

• Margin protection. AI inference is a cost line for the vendor (LLM tokens, retrieval indexing). Charging separately lets the vendor pass that cost through and isolates upside from the base subscription.
• Renewal lever. An unbundled AI line is the natural negotiation lever at year three. The base subscription holds; the AI upsell expands.
• Customer-specific value. Not every customer wants AI. Bundling it forces all customers to subsidise a feature only some will use; unbundling lets the buyer self-select.

• Which features are AI-powered, and which are conventional rules-based? Pin the difference; vendors increasingly label everything “AI” for marketing reasons.
• What is the underlying foundation model (Anthropic Claude, OpenAI GPT, Google Gemini, Meta Llama, vendor-trained)? What is the retention and training policy on customer data?
• Where does inference run (vendor cloud, customer cloud, region)? This is a data-residency question for EU/UK buyers.
• What is the per-user-per-month uplift for the AI tier or AI add-on? Insist on the line being itemised even if you take the bundle.
• What is the rate-limit, fair-use, or usage-overage policy? Some vendors apply soft caps on AI-generated content units.
• What does the AI feature actually do for a measured outcome (time-to-publish for an author, search hit-rate for a learner, completion lift for a path)? Push for evidence.

AI features earn their uplift when they address a real bottleneck. The two most reliable wins in LMSes today are authoring (SME time savings on first-draft content) and tagging (better search and recommendations on a large existing library). AI playlists and AI assessments are more variable and depend heavily on content quality. If your content library is small or your authoring volume is low, the AI tier may not return its cost. Pilot it for one quarter under the contract, with the ability to drop the line at renewal.