The AI Design Stack Just Reorganized — Moda at 99, Pencil at 75, Repaint at 71. The A.R.C. Breakdown.

Architecture (40%): Moda's bet is the strongest in the category — keep AI generation inside an editable canvas instead of producing a flattened export. That means generated assets stay reworkable: change a color, swap a layer, restructure the layout without re-prompting from scratch. The trade-off: the canvas layer adds learning surface. Operators expecting a one-shot generator find Moda heavier than Midjourney or DALL-E for the first 10 minutes; designers find it instantly more useful.

Reliability (35%): Solid for its age. The canvas state model is consistent (layers don't drift between sessions), generation latency is predictable, and the export surface is stable. The thin spots: occasional state-sync issues on long collaborative sessions, and undo history doesn't always span generation boundaries cleanly.

Context (25%): The +70 delta at viral 99 is the strongest single-tool signal in Design this week. Pencil and Repaint moving alongside (+50 and +53) suggest the category is rotating toward editable-output design tools as a class, not just toward Moda specifically. Moda is the leading edge of a real category shift, not a vanity spike.

Composite read: Moda is the strongest current pick for designers and brand teams that want AI generation without losing canvas control. Operators who want one-shot output should still reach for category-specific image tools — but for production design work, Moda is the right stack position.

Architecture (40%): Pencil is the most specialized of the three — it's built for ad-creative workflows specifically, with brand-kit ingestion, variant generation, and performance-feedback loops baked in. The architecture is correct for marketing teams running paid acquisition; it is the wrong architecture if you want general-purpose design work.

Reliability (35%): Mature. The brand-kit pipeline, variant rendering, and platform export integrations (Meta, TikTok, Google) all work the way you'd expect from a production ad tool. The performance feedback loop is the weakest link — the underlying signal quality depends on your ad spend and attribution setup more than on Pencil itself.

Context (25%): +50 delta at viral 75 is the strongest Pencil week in months. The signal: marketing teams are consolidating away from general-purpose generators (Midjourney, DALL-E) for paid creative, and toward tools that close the loop on creative performance.

Composite read: Pencil is the right tool for production paid-creative workflows. Outside that workflow, it's the wrong tool — its architecture is too specialized to be a generalist design stack pick.

Architecture (40%): Repaint's bet is depth in a single workflow — take an existing asset, apply a style direction, get a brand-consistent re-treatment. The architecture is narrow on purpose. For teams that need restyle volume (catalog refreshes, brand re-treatments, seasonal variants), the focused architecture beats generalist tools on output quality.

Reliability (35%): Good. Style consistency holds across batch jobs, brand-direction adherence is high, and the failure modes are visible (a bad restyle is obviously bad, not silently wrong).

Context (25%): +53 delta at viral 71 is strong sustained adoption for a narrow tool. The signal: restyle is finally being treated as a first-class workflow, not a side feature of a generalist tool.

Composite read: Repaint is the right tool for the right job — high-volume restyle work with brand consistency requirements. Pair it with Moda for the canvas work and Pencil for the paid-creative pipeline, and you have a real stack.