In 30 seconds
- Four team types: stream-aligned, platform, enabling, complicated-subsystem.
- Three interaction modes: collaboration, X-as-a-service, facilitating.
- Design teams around limited cognitive load.
Four team types
| Type | Purpose |
|---|---|
| Stream-aligned | Delivers value to customers end to end |
| Platform | Provides internal services so others move faster |
| Enabling | Helps other teams build new capabilities |
| Complicated-subsystem | Owns a part needing deep specialist knowledge |
Matthew Skelton and Manuel Pais introduced the model in Team Topologies (2019).
Three interaction modes
- Collaboration: working closely for a period
- X-as-a-service: one team consumes another's service
- Facilitating: one team helps another learn
Cognitive load
Teams can only handle so much. Give each team a scope it can fully understand, and use platforms to absorb complexity. Team boundaries shape system boundaries, as Conway's Law predicts; see microservices.
Myth vs reality
Myth · Every team should be stream-aligned.
Reality · Most should be; the other types support them.
Myth · Interaction modes are permanent.
Reality · They change as teams learn.
Myth · A platform team is an ops team renamed.
Reality · It provides self-service products for other teams.
The AI angle
AI agents and platforms are starting to look like new kinds of team members and services. Cognitive load still applies: someone must understand and own what agents produce.
FAQ
How many stream-aligned teams are typical?
Usually most of the organization's teams.
What's the inverse Conway manoeuvre?
Designing teams to get the system architecture you want.
How does this relate to the Spotify model?
Both organize around autonomy; Team Topologies focuses on flow and interactions.