Full example council

Serverless vs. containers.

A public replay of the Model Council workflow: prompt, independent answers, critique, stance changes, and a final decision artifact.

Prompt

Compare a serverless architecture against a container-based architecture for this product.

Round 01

Independent answers

The models split between serverless-first, conditional hybrid, and container-first positions.

4/5

Position

Serverless first for early or spiky products

4/5

Position

Conditional, with hybrid as a common end state

4/5

Position

Pick based on scaling, cost, performance, velocity, and lock-in

4/5

Position

Initially container-first for non-trivial products

Verdict table

The strongest result was not a single model winning. It was the council narrowing the decision to constraints that matter.

Core agreement

Serverless is strongest for early, spiky, short-lived workloads.

Key disagreement

Containers win only when sustained traffic, latency, or runtime control justify operations.

Useful refinement

Serverless containers are the pragmatic bridge between functions and full Kubernetes.

Serverless is strongest for spiky traffic, small teams, short-lived tasks, and early-stage products because it reduces operations and scales to zero.

Containers win for sustained high-throughput workloads, latency-sensitive systems, long-running jobs, stateful services, and strict runtime control.

Consensus is not truth. A council run makes disagreement, assumptions, and decision criteria visible.