The opening chapter
Chapter I is where every project begins its life as a document. The Direction opens with your original concept — the raw idea, preserved exactly as you wrote it — followed by the house’s campaign summary, the key values of the direction, the models you selected, and, when the Engine has found one, a directional conflict panel.
Keep the original concept in view as you work. It is the sentence everything else must answer to, and it is deliberately kept verbatim: when a project drifts, the fastest diagnosis is to reread what you actually asked for.
Reading the summary
The campaign summary is the house restating your idea in its own words. It is the same move a good agency makes at the top of a treatment: here is what we heard, here is the world we intend to build from it. Read it slowly, because it is the earliest and cheapest place to catch a misunderstanding.
If the summary describes your campaign more sharply than your brief did, take the gift — that sharpness will already be in the Blueprint and the prompts. If it describes a campaign you would not sign, do not push forward hoping the frames will drift back toward your intention. They will not. Revise the brief and regenerate; the summary is your contract with the Engine, and you should only proceed on a contract you would put your name to.
The key values and the models
Beneath the summary sit the direction’s key values — brand, campaign type, output type, realism level, and the desired emotion where you set one. These are the load-bearing decisions of the brief laid out as a ledger, so a collaborator can understand the project’s spine in ten seconds without reading a paragraph.
The selected models follow as a row of plates. They record which architectures this direction was composed for — each one received its own native prompt set. When you later change the selection in the Studio and regenerate, this panel is where the change shows its face.
When the brief argues with itself
A directional conflict is the Engine finding a genuine contradiction inside your direction — an emotion that fights the campaign type, a format that undermines the intended register. Most tools resolve such tensions silently, averaging both intentions into something that serves neither. The Engine names the tension instead, in plain language, in an amber panel you cannot mistake for decoration.
This is a feature of the house, not an error state. A brief with a visible tension is a brief that contained two real ideas; the panel exists so that you, not a sampling algorithm, decide which one leads.
Each conflict arrives with resolution paths — two or more concrete directions the campaign could commit to, each described in terms of what it would change. They are not a menu of compromises; they are forks.
Resolving a conflict is a creative decision
Treat the resolution the way you would treat a fork in a treatment meeting. Read both paths aloud. Ask which one the brand would recognise itself in, and which one merely sounds interesting. Then choose, revise the brief in the direction of the chosen path, and regenerate — cleanly, without smuggling half of the abandoned path back in.
The discipline here is to choose once and mean it. A campaign built on a resolved tension has a point of view; a campaign built on an unresolved one has a hedge. Clients can tell the difference even when they cannot name it, and so can the frames.