Customer Stories

Power your GenAI experiments with a library of full page screenshots of your product

How Metadata accelerated GenAI experiments using screenshot automation to continuously capture updated screenshots across their SaaS application.

Company
Metadata is a demand generation platform that uses AI agents and automation to optimize paid advertising for b2b brands, streamlines audience targeting, and enhances campaign experimentation—allowing marketers to focus on strategy and creativity.
Industry
Software, AI, Advertising
Key Features
Login Profile, Recipe Builder, Custom Rendering, Export to storage bucket, and Implementation Services.

Overview

As the demand for Generative AI (GenAI) continues to rise, companies are constantly seeking high-quality training datasets to enhance AI-driven insights and automation. LaunchBrightly serves technical writers and knowledge managers with a screenshot automation platform that automatically and continuously keeps their help documentation updated with fresh and accurate product screenshots. And we are now seeing exciting new, inventive use cases for the technology emerge—particularly in AI-driven research and development.

For Metadata, an AI-powered marketing platform, staying ahead of the curve meant exploring GenAI experiments that required a large-scale, high-fidelity dataset of screenshots from across every route of their SaaS application. But gathering and maintaining such a dataset manually posed an operational challenge.

Challenge

AI agents can truly become one of your greatest competitive advantages as an organization. However, the key to unlocking the potential of these advanced GenAI tools often lies in an organization's ability to ground these tools in the organization's unique content, and leverage organization-specific data sets.

Getting a large, fully exhaustive pool of application screenshots manually is a massive challengeGil Allouche, CEO of Metadata

The team at Metadata were embarking on a set of experiments to explore what GenAI could do for the frontend engineering team. To make this experiment meaningful, they needed a vast collection of high-resolution, lossless screenshots from across their application—critical for enabling GenAI to analyze UI patterns, structure, and potential areas for automation and optimization.

Manually gathering and maintaining hundreds of full-page screenshots—especially in a rapidly evolving SaaS environment—was an overwhelming task. “Getting a large, fully exhaustive pool of application screenshots manually is a massive challenge. And if we wanted to update all of them again just a few weeks later, we'd drown,” says Metadata CEO, Gil Allouche.

To fuel their Gen AI experiments, they needed a fully automated, scalable solution—one that could reliably capture and update a complete library of SaaS screenshots, without manual effort.

Solution

To overcome these challenges, Metadata partnered with LaunchBrightly, leveraging their team of Implementation Specialists and screenshot automation platform to continuously and automatically capture screenshots from behind the applications login, and deliver a library of screenshots at scale.

LaunchBrightly's team of experts set about:

  • Configuring a secure Login Profile to allow for seamless programmatic capture of screenshots from behind the authentication of Metadata’s application;
  • Identified every key route in Metadata's application, and systematically created screenshot automation recipes to repeatedly capture full page screenshots of each of these key routes;
  • Configured the screenshots to be captured in high-resolution (retina) with lossless compression, ensuring maximum fidelity—critical for use in advanced GenAI where every pixel matters; and
  • Established a storage integration to an external AWS S3 bucket, allowing Metadata to easily export its library of automated screenshots on an ongoing basis, and giving them full control over dataset delivery.

“Seeing LaunchBrightly quickly set up automation for all our views and just run, and rerun, screenshot captures in lossless quality was priceless. It helped increase the velocity of our GenAI experiments to no end. In addition - the customer support was superb” said Gil.

The Outcome

By leveraging the expertise of LaunchBrightly's Implementation Specialists, and deploying its screenshot automation platform, Metadata now has in place a fully automated, scalable solution for its screenshot capture. In the matter of just a few weeks, they have built a large-scale, high-fidelity and dynamic repository of retina–quality screenshots.

The team at Metadata are now set to accelerate their GenAI efforts with:

  • Comprehensive coverage: with hundreds of full page screenshots capturing every route in the SaaS application;
  • Embedded metadata for AI optimization: with route metadata injected in each individual image file for classification and use in GenAI experimentation; and
  • Automatic and continuous updates: each of Metadata's screenshots is able to be reprocessed at any time—including in mobile or dark mode variants—without additional setup, ensuring that AI models always train on the most up-to-date visuals of their application.

“Being able to rerun these in other modes (resolution, language, dark, etc), should we choose to do so, is magical” Gil says, “Especially since we’re still experimenting with what render is best for our Gen AI work.”


Need a bucketful of high-resolution (retina) screenshots of your app for an experiment? Drop us a line at hello@launchbrightly.com — or snag a time on our calendar. 🤗

.