PathwayAI compared with Udemy and Coursera
A point-in-time comparison on the dimensions where PathwayAI actually differs. Every PathwayAI claim names the code that backs it; the architecture page has more detail.
How they differ
#| Dimension | PathwayAI | Udemy | Coursera | Sources for Udemy and Coursera |
|---|---|---|---|---|
| Citation model | Every RAG chat answer cites the exact source passage and page it came from, and refuses to answer beyond what the evidence supports. (see lib/rag/prompt.ts) | Our reading, not checked: No citation model — course Q&A and reviews are not grounded against specific source material. | Our reading, not checked: No citation model — course content is presented without per-claim source attribution. | Not checked against vendor material. Treat the Udemy and Coursera cells in this row as our reading. |
| Skill-gap-driven roadmap | Generates a personalized roadmap from an embedding-based comparison of your documents against a skill taxonomy, not a static catalog browse. (see lib/competency/extract.ts) | Our reading, not checked: Catalog search and manual browsing by category/keyword; no automated skill-gap extraction from a learner's own documents. | Our reading, not checked: Catalog search with some skill tagging on courses, but no automated per-learner skill-gap extraction from uploaded documents. | Not checked against vendor material. Treat the Udemy and Coursera cells in this row as our reading. |
| Verifiable credentials | PDF exports can carry an HMAC-signed, expiring QR code resolving to a public verification page. (see lib/reports/verification.ts) | A completion certificate has its own URL that can be shared and opened by others. Udemy's help pages do not describe a cryptographic signature that can be checked offline. | A certificate has a public, non-expiring verification URL that anyone can open. Coursera's help pages do not describe a signed token embedded in the document itself. |
|
| Skill taxonomy structure | Skills, offers, and roles are derived into a real, queryable knowledge graph with clusters and weighted edges — metrics are computed from the graph, not hardcoded. (see lib/graph/build.ts) | Our reading, not checked: Courses are organized by category and tag; no derived graph of skill relationships. | Our reading, not checked: Courses are organized by subject and specialization; no derived graph of skill relationships. | Not checked against vendor material. Treat the Udemy and Coursera cells in this row as our reading. |