Digests

7h09 of launch event video is a lot. Each digest keeps the features that matter for one audience, grouped by area and deep-linked to the second where they explain it, with the rest as "also worth a look".

Developers

If you write AL, build extensions, or own a partner's dev tooling.

108 min to watch · 67 features in the playlist · 112 more worth a look

Consultants

If you implement, configure and train: finance, supply chain, e-documents, sustainability, Expense Agent, reporting, Copilot and the integrations.

134 min to watch · 49 features in the playlist · 171 more worth a look

Administrators

If you run the tenants: admin center, updates, permissions, security, environments, telemetry.

64 min to watch · 51 features in the playlist · 60 more worth a look

Decision makers

If you decide what to roll out, what to budget and what to tell the business.

96 min to watch · 24 features in the playlist · 74 more worth a look

How the lists are made

Developers: The extraction model rated every feature's developer relevance (high, medium, low). High is the playlist, medium is also worth a look.
Consultants: Functional areas only, features that were demoed on stage for at least a minute and a half, minus the ones rated high developer relevance. The rest of the functional areas is also worth a look.
Administrators: Everything in the admin and platform area, plus features elsewhere tagged security, permissions, updates, sandbox, telemetry, MDM, operations or PTE management (at least half a minute of airtime, not rated high developer relevance). Features tagged admin or setup are also worth a look.
Decision makers: The agent and Copilot features that got a real demo (two minutes or more, outside the developer tooling), plus the two longest features of every area so you see what each team spent its stage time on. Also worth a look: everything announced for later (the roadmap), everything in preview, and the shorter agent and Copilot items.

The rules live in config/audiences.json. The developer rating comes from the extraction model; the other three are plain rules over area, tags, demo and airtime. Blunt by design: change the file, rebuild, and the digests follow.