Three frameworksFive stages, no score bands
Agents do the work. One human owns the call. This is the model.
Three assessment frameworks, one ladder, and no scores. Each framework asks a single question that a non-technical executive can answer honestly in about a minute.
A number invites debate about the number. A description invites recognition. Every stage here is written so that the right one is obvious and slightly uncomfortable.
Three questions separate a team with agents from a team that has an operating model.
Each one is deliberately answerable without a tool, a survey, or a consultant. If answering any of them requires a project, that is already the finding.
The judgment boundary
Where does agent work stop and the call begin, and does every call have a name on it?
A written line per stage of work, and the owner of each approval.
If nobody can name the owner of last quarter’s biggest call, you are at stage one.
The landing read
Is every launch read the same way, including the ones leadership championed?
One definition of landed, watch, and stalled, applied without exception.
If your favourite project is the one exception, you do not have a definition.
The decision ledger
Was the belief written down before the data arrived, and revisited after it did?
A dated record of what you expected, next to what actually happened.
Without it, every retrospective is a memory contest and memory always flatters.
Five stages, described rather than scored, so the honest answer is the easy one.
Read the five descriptions and pick the one your last quarter would support in front of a sceptic. Most organisations I have read sit at stage two and describe themselves at stage four.
- Stage 1 of 5
Effort is the report
The review is a list of what shipped. Nobody asks what changed.
- Stage 2 of 5Most teams read here
Somebody asks whether it landed
The question gets asked, answered with a story, and not asked again.
- Stage 3 of 5
The question is asked the same way every time
One definition of landed, applied to every launch, including the ones you liked.
- Stage 4 of 5
Agents do the reading, a person makes the call
The evidence assembles itself. The judgment still has one name on it.
- Stage 5 of 5
The answer changes what gets funded
Work that did not land loses its budget without anyone needing to make the case.
The line between agent work and the call does not move as you climb.
What changes between stage one and stage five is how much evidence arrives before the call, and how consistently the call gets read afterwards. The call itself stays exactly where it is.
- Business fit
- Design quality
- Technical fit
- Final approval
Judgment. Never delegated, never automated.
- Read the context
- Draft the options
- Check the rules
- Propose the next move
Volume. Fast, tireless, and never the owner.
Everything left of the rule has a name attached to it.
Each framework has a page behind it that shows the work.
The eight claims behind the model
Why building got cheap, what got scarce, and the five-step loop that follows from both.
The rituals that survive the change
Which meetings were asking a real question, and which were asking whether anyone was busy.
What you say in sixty seconds
The four questions a board asks, and how each stage of the ladder changes your answer.
Now watch the model run on a real company.
Throughline applies all three frameworks to one company's public record and produces a stage read you can disagree with, which is the only kind worth having.