ENG-001 // K-12 // BC INTERIOR
Provincial science strand refresh
A regional school district needed forty modular units for Grade 9–10 physical sciences aligned to BC curriculum codes. Legacy PDFs were scattered across three predecessor textbooks. We ingested the approved corpus, built a prerequisite graph covering 112 concepts, and generated draft modules with hands-on lab prompts.
District subject leads reviewed batches of five modules per week. Factual errors caught at review dropped from a typical vendor baseline of 14% to under 2% by week six, once the graph constraints matured. Final SCORM packages deployed in Brightspace before the September intake.
Duration: 14 weeks | Scope: Generative Content Engines + Assessment Generation
ENG-002 // POST-SECONDARY // LOWER MAINLAND
College nursing RAG pilot
A community college nursing faculty wanted a query tool that answered student questions using only their program handbook, clinical guidelines, and past exam blueprints — never the open web. We deployed a private retrieval stack in a Canadian cloud region with role-based access for students, instructors, and administrators.
Citation accuracy exceeded 94% on a 200-question faculty audit set. The pilot ran one semester with weekly bias reviews covering cultural competence scenarios. Governance documentation satisfied the college's accreditation self-study without additional vendor calls.
Duration: 10 weeks build + 16 weeks pilot | Scope: Curriculum RAG + Responsible AI Governance
ENG-003 // CORPORATE L&D // NATIONAL
Financial compliance training overhaul
A federally regulated employer faced annual compliance training that staff clicked through without retention. We replaced static slides with adaptive pathways keyed to role family — front-line, supervisory, and executive tracks. Checkpoint items were generated from their internal policy corpus with legal review gates at each score band.
Post-deployment analytics showed a 31% reduction in time-on-task for staff who passed first attempt, and a 22-point lift in follow-up knowledge checks versus the prior vendor content. The client retained full ownership of all generated assets and graph structures.
Duration: 18 weeks | Scope: Adaptive Learning + Learning Analytics + Assessment Generation
ENG-004 // PUBLISHER // PAN-CANADIAN
Mathematics supplement item bank
An educational publisher required 800 formative items spanning Grades 6–8 mathematics with French and English stems. We built parallel generation pipelines sharing a single concept graph to keep translation alignment tight. Accessibility metadata met WCAG 2.1 AA for every item.
Psychometric sampling on a 120-item subset showed acceptable difficulty spread. The publisher integrated the bank into their existing authoring tool via API. Ongoing governance advisory covers model updates quarterly.
Duration: 22 weeks | Scope: Assessment Generation + Responsible AI Governance
Your challenge belongs on this list
We scope pilots honestly. If your problem is outside our practice areas, we will say so and point you elsewhere.
Bring us a content challenge
