01 / 05 · Brief
Tell us what you need. We hand you the plan.
Most AI tools start at the prompt. Axion Studio starts at the brief. You describe what you need — chat or form — and the Asset Dependency Planner returns a stage-by-stage plan with explicit approval gates at every layer. Nothing generates until you say go.
The model
Gemini 2.5 Pro decomposes your ask into 5 layers
The planner doesn't just expand your prompt — it breaks the work into a directed acyclic graph: brand identity (logo, colors, LUT, voice) → templates (PDP hero, lifestyle, ad creative) → substrates (the raw renders) → composites (logo + text + LUT applied) → distribution (sized, signed, channel-pushed).
Each stage carries an approval gate: per_stage, per_batch, per_asset, or none. Tinder-grid review for batches; full-detail review for hero shots.
Category-agnostic
Same code path for any vertical
The planner reads your brand's product list and adapts. Peptide, cosmetics, apparel, food, electronics, B2B SaaS — same DAG structure, different content.
If your product list is missing or incomplete, the planner returns a partial plan with questions_for_customer instead of guessing. You answer; it re-plans.
API
The planner is a public endpoint
Pipe briefs in from anywhere — Slack, Notion, your PIM. The planner returns a validated DAG (cycles raise, dangling deps strip).
POST /agents/plan-assets
{
"brand_id": "your-brand-handle",
"brief": "Spring campaign for new product launch — hero shots,
3 lifestyle variants, 1 talking-head 30s, email
graphics for the welcome sequence."
}
→ AssetPlan { stages: [...], gates: [...], questions_for_customer: [] }By the numbers