Robinhood / AI systems · Creative operations
An AI creative dashboard, a custom Figma integration, and automated disclosure insertion for a regulated advertising workflow.
The challenge
Turn campaign briefs into ad creative variations with disclosure insertion and human review, and connect generated assets to an internal platform that had no API.
My contribution
The outcome
My résumé reports a 10× increase in ad output from the workflow, which combines disclosure insertion with human review. The work connected generation with the practical requirements of asset delivery and regulated copy.
My ownership
Through Buildroom.ai, I built the creative dashboard, custom Figma integration and disclosure-insertion workflow that connect a campaign brief to reviewable creative.
The wider team & tools
This was client work through Buildroom.ai. The client’s campaign direction, brand assets and approval decisions remain theirs. Generated copy and inserted disclosures still require human review.
How it fits together
Campaign inputs become structured headlines, subheads and calls to action.
Claude generation, Google Sheets records and a dashboard for review and feedback.
Reusable layouts combine copy, brand assets and a dedicated disclosure field.
A custom integration bridges the final handoff described in the project scope.
The archive contains a Next.js app, server-side model and Sheets integrations, and Figma Plugin API templates. The delivery connection is described in my résumé.
Engineering & product judgment
The app has brief intake, review, variation and performance views. Regenerating copy saves feedback and clears its previous approval state.
Generating a plausible line is only one step. A reviewer needs to understand what changed and make the approval decision.
Figma templates accept a dedicated disclosure field, so the layout carries that text alongside the headline and visual.
Insertion is a useful control, but it does not establish legal accuracy or replace the client’s review process.
The Figma implementation loads fonts once, caches logo variants, creates multiple layouts in a batch and cleans up temporary assets afterwards.
Batching reduces repeated setup work. Batch size still needs to fit the host’s execution limits; no new latency benchmark is claimed here.
An iteration from the archive
The Figma implementation notes describe artwork whose outer frame contained empty space. Scaling that frame did not scale the visible illustration as expected.
The template workflow uses the inner artwork bounds and calculates a placement zone between the text and logo, with padding. Related helpers handle contrast changes for darker backgrounds.
Creative automation needs layout-aware checks as well as correct text. Small integration details decide whether the output is actually usable.
Outcomes, with context
Résumé-reported workflow result. Baseline volume and measurement period are not included in the source.
Square, story, portrait and landscape templates verified in the project archive.
Disclosure insertion and review tooling support the process; they are not a guarantee of regulatory compliance.
Ownership and 10× outcome: September 2026 résumé. Workflow, approval reset and batching details: original app and Figma source. Actual app screenshots are available in the related build.