MarTech Retention Platform Cuts Figma-to-Campaign Design Time from Hours to Minutes with Claude on AWS Bedrock
At a glance
Subscription retention platforms live or die by how fast growth teams can launch new in-app prompts, but turning a Figma design into a production-ready pop-up for streaming, media, and publishing merchants has always required scarce developer time, capping how many campaigns and A/B tests a merchant can run in a quarter.
NMD built and productionized an AI pipeline on AWS Bedrock that reads Figma design files and generates production-ready platform objects directly, cutting a manual build-and-QA cycle down to about two minutes per design across three funded engagement phases.
This subscriber engagement platform’s growth teams can now turn a finished Figma design into a live, testable retention prompt without filing an engineering ticket, opening the door to a self-serve, conversational AI design layer.
Industry
Use Case
Workflow Automation, Multimodal Content Extraction, Agentic AI
Solution implemented
- Amazon Bedrock with Claude models power multimodal analysis of Figma layouts and typography
- AWS Lambda proxies Figma API calls, bypassing IP-blocking issues at the network edge
- Amazon ECS Fargate runs the translation pipeline as a containerized, horizontally scalable service
- Amazon DynamoDB tracks granular job status for asynchronous, pollable pipeline execution
- Per-company Amazon S3 buckets isolate design assets with clean access control boundaries
- Pipeline achieves approximately 90% visual fidelity between Figma source and platform output
The value equation
- Growth teams launch new retention prompts in minutes instead of waiting on engineering tickets
- AI design pipeline removes a recurring developer bottleneck from every new campaign and test
- Frees engineering capacity for platform features instead of one-off Figma-to-UI translation work
- Positions the platform's AI translation layer as a differentiator against competing engagement tools
- Establishes a reusable Bedrock pipeline ready to extend into generative, chat-driven design tools
Company Snapshot
A real-time subscriber engagement platform, part of a larger subscription management suite, that powers personalized retention prompts for streaming, media, and publishing brands.
Location
US (West Coast)
Customer Situation
This platform’s merchants, subscription brands across streaming, media, and publishing, rely on real-time, personalized prompts to reduce churn and drive upgrades. New prompt designs originate in Figma, but turning those designs into live, production platform objects required manual developer translation for every layout, color scheme, and responsive variant.
That translation step throttled how quickly growth and marketing teams could test new retention offers. This design-to-production bottleneck is common across MarTech and CRM platforms that separate design tools from the systems marketers actually operate in.
NMD Solution
NMD reviewed the Figma-to-platform workflow and found the repeated manual translation step, not campaign strategy, as the real constraint on growth team velocity. The solution required a multimodal AI pipeline on Amazon Bedrock using Claude models to parse Figma node data, visual layout, and styling, then generate platform-ready popup objects, backed by AWS Lambda, ECS Fargate, DynamoDB, and S3 for secure, scalable production deployment.
What We Delivered
NMD delivers a production AI translation pipeline that converts Figma-designed pop-ups and banners into live platform objects in one to two minutes, with support for OAuth, plan, and personal Figma tokens, asynchronous job polling, and per-company asset isolation. The pipeline runs in the client’s own AWS environment, and the client’s product team is scoping a Q3 follow-on to add conversational, AI-generated design edits directly inside the platform.
Ready to Transform Your MarTech Business?
See how New Math Data can transform your MarTech business with AWS-powered innovation.