What do provider claims leave for educators to decide? Examine possible uses, the work of checking them, and the choices available to students and instructors.
Each profile separates a dated provider statement from an original instructional interpretation. Announced capability, account availability and evidence of learning are different questions.
Read the comparison frameworkCompare the directions
6 companies
Persistent agent sessions, dynamic workflows, and explicit execution controls
Longer-running delegation makes intermediate decisions and retained context part of the teaching design. More agents do not create independent evidence.
Latest selected development: 2026-10-20. Sources checked 2026-10-24.
Models, tools and retrieval
A model-and-API route into tool-using systems, with dated updates that matter to reproducibility.
Latest selected development: 2026-10-17. Sources checked 2026-10-24.
Delegation inside Workspace, multimodal inspection, and persistent work agents
Course materials already live across documents, slides, video, and communications. Connecting those locations makes source selection and instructional coherence more consequential.
Latest selected development: 2026-10-09. Sources checked 2026-10-24.
Personal agents and deployable models
Personal-agent products and model releases raise different questions about authority, persistence and deployment.
Latest selected development: 2026-10-15. Sources checked 2026-10-24.
Institutional tools and learning support
Preparation, student practice and institution-managed work appear within a broader education ecosystem.
Latest selected development: 2026-05-19. Sources checked 2026-10-24.
Managed execution, browser actions, and finished artifacts
The relevant change is the unit of delegation: an agent can work across tools toward a reviewable artifact. The teaching question is which intellectual decisions remain visible.
Latest selected development: 2026-10-16. Sources checked 2026-10-24.
No company matches. Clear the search or try another term.
| Company | Selected direction | A question for the instructor |
|---|---|---|
| Anthropic | Persistent agent sessions, dynamic workflows, and explicit execution controls | Can reviewers reconstruct why a persistent task changed a course example? |
| DeepSeek | Models, tools and retrieval | Which errors survive instructor review when a workflow searches and drafts? |
| Delegation inside Workspace, multimodal inspection, and persistent work agents | Which announced functions are actually enabled for the institution’s account and region? | |
| Meta | Personal agents and deployable models | What does a learner infer about intention and trust from a personal-agent interface? |
| Microsoft | Institutional tools and learning support | What reasoning do students perform before receiving a hint or answer? |
| OpenAI | Managed execution, browser actions, and finished artifacts | Which errors remain invisible to an instructor who is not already expert in the topic? |